7027953_Method_and_system_for_diagnostic

7027953_Method_and_system_for_diagnostic - USOO7027953B2...

Info iconThis preview shows pages 1–58. Sign up to view the full content.

View Full Document Right Arrow Icon
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 2
Background image of page 3

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 4
Background image of page 5

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 6
Background image of page 7

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 8
Background image of page 9

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 10
Background image of page 11

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 12
Background image of page 13

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 14
Background image of page 15

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 16
Background image of page 17

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 18
Background image of page 19

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 20
Background image of page 21

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 22
Background image of page 23

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 24
Background image of page 25

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 26
Background image of page 27

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 28
Background image of page 29

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Background image of page 30
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: USOO7027953B2 (12) Unlted States Patent (10) Patent No.: US 7,027,953 B2 Klein (45) Date of Patent: Apr. 11, 2006 (54) METHOD AND SYSTEM FOR DIAGNOSTICS 5,223,207 A 6/1993 Gross et 31. AND PROGNOSTICS OF A MECHANICAL 5,239,468 A 8/ 1993 SeWGTSkY et 31~ SYSTEM (Continued) (75) Inventor: Renata Klein, Misgav (IL) OTHER PUBLICATIONS . . _ . Han, Y. et al., “A selfileaming expert system for grinding (73) Ass1gnee. RSL Electronlcs Ltd., M1gdal Haemek Vibration diagnosis,” Dept of Prod. Syst. Eng” Toyohashi (IL) Univ. of Technol., Japan, Conf' Artificial Intelligence in the ( * ) Notice: Subject to any disclaimer, the term of this Paczfic le’ Procéedings of tile Paczfic le Imemanonal . . Conference on Artificial Intelligence, p. 146751, 1991. patent 1s extended or adjusted under 35 cc - - - - - U S C 154 b b 227 d Carr, H.R., Aeroiengme V1brat1on d1agnos1s on a DSPiac- ’ ’ ’ ( ) y ays' celerated personal computer,” Dept. of Exp. Vibration, RollsiRoyce plc, Bristol, UK; Conf' IEE Colloquium on (21) APP1~ N05 10/3343477 Advanced Vibration Measurements, Techniques and Instru— 22 F] d: D _ 30 2002 mentation for the Early Prediction ofFailure, Digest No. ( ) le ec ’ 105, p. 5/1416, 1992. (65) Prior Publication Data Arefzadeh, S. et al., “Diagnosis of diesel engines based on Vibration analysis and fuzzyilogic,” lnst. Fur Bergwerk- Us NOS/0096873 A1 May 5’ 2005 siund Huttenmaschinenkunde, Aachen Univ. of Technol., (51) Int, Cl, Germany; Conf' Fourth European Congress on Intelligent G06F 15/00 (2006.01) Techniques and Soft Computing Proceedings, EUFIT ’96, Part vol. 2, p. 14977500, 1996. (52) us. Cl. ..................... .. 702/184; 702/182; 702/183; Cempel, C, “Theory of energy transforming systems and 702/185 their application in diagnostics of operating systems,” Inst. (58) Field of Classification Search ................. .. 702/54, ofAPP1~MeCh~sTech~UniV~P0ZHans PolandsAPPliedMa’h- ematics and Computer Science, V01. 3, 3, 700/279, 280, 287; 290/408; 701/120; 244/1 N; 1993 . . 340/945; 706/52 Primary ExaminergCarol S Tsai See app11cat1on file for complete search h1story. (74) Attorney] Agent] or Firm73kaddena Arps, Slate, Meagher & Flom LLP (56) References Cited U.S. PATENT DOCUMENTS 3,758,758 A 9/1973 Games et 31. 4,283,634 A * 8/1981 Yannone et al. ........ .. 290/40 R 4,320,662 A 3/1982 Schaub et al. 4,426,641 A l/l984 Kurihara et al. 4,488,240 A 12/1984 Kapadia et al. 4,787,053 A * 11/1988 Moore ...................... .. 701/123 4,985,857 A 1/1991 Bajpai et al. 4,988,979 A * 1/1991 Sasaki et al. ............. .. 340/683 4,989,159 A 1/1991 Liszka et al. 5,150,855 A * 9/1992 Kaptein ................... .. 244/1 N 5,210,704 A 5/1993 Husseiny 5,211,539 A * 5/1993 McCarty .................... .. 416/61 (57) ABSTRACT A Vibrational analysis system diagnosis the health of a mechanical system by reference to Vibration signature data from multiple domains. Features are extracted from signa- ture data by reference to pointer locations. The features provide an indication of signature deviation from a baseline signature in the observed domain. Several features appli- cable to a desired fault are aggregated to provide an indi- cation of the likelihood that the fault has manifested in the observed mechanical system. The system may also be used for trend analysis of the health of the mechanical system. 47 Claims, 8 Drawing Sheets pwnnw ENGINE US 7,027,953 B2 Page 2 US. PATENT DOCUMENTS 5,951,611 A 9/1999 La Pierre _ 5,978,727 A 11/1999 Jones et a1. 5,293,585 A 3/1994 Morita 6,004,017 A 12/1999 Madhavan 5,351,247 A 9/1994 DOW etal. 6,014,447 A 1/2000 Kohnen et a1. 5,383,133 A 1/1995 Staple 6,014,598 A 1/2000 Duyar et a1. 5,519,637 A 5/1996 Mathur ..................... .. 700/280 6,027,239 A >x< 2/2000 Ghassaei .................. N 700/279 5,526,292 A 6/1996 Hodgson et a1. .......... .. 700/280 6,175,787 Bl 1/2001 Breed 5,551,649 A 9/1996 Kaptein ................... .. 244/1 N 6,192,325 B1 2/2001 Piety et a1. 5566092 A 10/1996 Wang etal. 6,199,018 B1 3/2001 Quist et a1. 5,574,646 A 11/1996 Kawasaki et a1. 6,260,004 B1 7/2001 HayS et 31. 5,579,232 A 11/1996 Tong etal. 6,295,510 B1 9/2001 Descenzo 5,600,576 A * 2/1997 Broadwater et a1. ...... .. 702/187 6,298,308 B1 10/2001 Reid et 31. 5,602,757 A 2/1997 Haseley etal. 6,301,572 B1 * 10/2001 Harrison .................... .. 706/52 5,610,339 A 3/1997 Haseley etal. 6,321,602 B1 11/2001 Ben-Romdhane 5,698,788 A 12/1997 M01 et 31~ 6,343,251 B1 1/2002 Herron et a1. 5,726,891 A 3/1998 Sisson et al. 6363303 Bl 3/2002 Benness 5,768,124 A * 6/1998 Stothers et a1. ............. .. 700/38 6,711,523 B1 >x< 3/2004 Bechhoefer et 31. ...... N 702/181 5774376 A 6/1998 Manning 2001/0001851 A1 5/2001 Piety et a1. 5,804,726 A 9/1998 Geib etal. 2002/0013635 A1 1/2002 Gotou et a1. 5,806,011 A 9/1998 Azzaro etal. 2002/0032544 A1 3/2002 Reid et a1. 5,809,437 A 9/1998 Breed 2002/0038199 A1 3/2002 Blemel ..................... .. 702/183 5,852,793 A 12/1998 Board fit 31~ 2003/0083794 A1 5/2003 Halm et a1. .. .. 701/29 5,854,993 A 12/1998 Grlchnlk 2004/0003318 A1 1/2004 Felke et a1. ................. .. 714/25 5,893,892 A 4/1999 Loefiler 5,895,857 A 4/ 1999 Robinson et a1. * cited by examiner US 7,027,953 132 Sheet 1 of 8 Apr. 11, 2006 U.S. Patent _ «Sui mdo: I v.00: I , 9-8: .1. #8: I 353533: me _ no mo «‘0 No Q m ZH/ZE) . __ ‘ m w“. H m 2.3.5. .015: a. v.00: U.S. Patent Apr. 11,2006 Sheet 2 0f 8 US 7,027,953 B2 FLIGHT DATA ENGINE DATA 115 ~> mama-Eu; """"" “ ‘ T-HUMS AIRBORNE UNIT +— 200 RS422.TCOP/IP DATA TRANSF ER Y UNIT (pm) 6300 \/ GROUND 5mm»: Fi ure 3 U.S. Patent Apr. 11,2006 Sheet 3 0f 8 US 7,027,953 B2 » PW4000 ENGINE 112 VIERATION <— 111 s NSOPS ” (NEW, T—HUMS , ENGINE 220 210 INSTALLED ‘1' I 230 EQUIPMENT 116-» E "E e 301 CONTROL PROCESSING mm ’ _ "M L DATA STORAGE 24 AIRCRAFT GSM 250—> COMMUNICATION COMMUNICATION ImERFACE CHANNEL GSM T ANTEMA " ‘ "'—*‘— “114"” "401"—“"_ “ W FLIGHT 117—) AND +— 113 12293:; A/C DATA EQUIPMENT GROUND BASED EQUIPMENT GROUND DIAGNOSTIG / WORK STATION .. , » , ESM GSM DATA TRANSFER V (_ 400 UNIT (DTU) Fi ure 4 US 7,027,953 B2 Sheet 4 of 8 Apr. 11, 2006 U.S. Patent near—mooom Emflan. £222: 9:53. ES: Eozficau Bum EV Eug BEE: Essa mmSEQEQ ngfiuoou Enema.“ Eu 9 azuow mEmmwooE ficmfi EmEoSmaos— ,ozgmwmuzmzu omv 33:52 fitmm xofmwoxofimzwiflea Eozmfiozcmg flea—~59. 52:932.: saun— mquE 0.5—3“— flaw. m 925 n 939“. > baggy US 7,027,953 B2 00 3m f 0 5 t 4 e 1 .||\ m m; /[ 20 I _ 7r, / 6 , 0 , o w 2 , $33.3: 73:395. 625 an: ,1:\I/ :.!,!_ n, + «2..qu: . 3:050: 3:20.: 2:95.! W i ASE! . , 96w d aux—.23 r \L\ \\ 0.. 8m :w I , A A x \ $743 Emu BE gum a 22.9.2.3 \ row U.S. Patent US 7,027,953 B2 Sheet 6 of 8 Apr. 11, 2006 U.S. Patent £925! .uuaciaxm _ .mzm do , "meEoE fi , .35.. :5 39.395.292.51 £21 do “25:52 a Own b whim?“— $320.:me 9.2.23qu 20 K 7“ :022 83 . L v r \ U.S. Patent Apr. 11,2006 Sheet 7 0f 8 US 7,027,953 B2 \ Wigner v Ville rpm Background spectrogram >-.\. 3-! a \ :s s s \ 3 2 -‘= / o— 2 LE __J9/ m i”! E O S a A #_ /~~\ E Q a . I: 96 L \fi/ ‘1 «pa, 5 ! “R? S 1 ‘5: ‘ —~—— 2 9 s g a g (n , _ US 7,027,953 B2 Sheet 8 of 8 Apr. 11, 2006 U.S. Patent EBRm 5qu m c _Emo...:mm m:_Emw:E_ow swim 25>: Swim 25>... wuamuflw goo—m k 22823 22:35 0:9» ucwfi. 5mm 30 :23. term a a 3553 be £305 Ems. 0mm omm Ema mam. “o swam: w " mmo E5» 9.04 AI :5. US 7,027,953 B2 1 METHOD AND SYSTEM FOR DIAGNOSTICS AND PROGNOSTICS OF A MECHANICAL SYSTEM FIELD OF THE INVENTION The present invention relates generally to a system for monitoring the condition of a mechanical system such as an aircraft engine. More specifically the invention relates to a diagnostic and prognostic method and system applicable to rotating machinery and in particular, aircraft engines and helicopter gearboxes. BACKGROUND OF THE INVENTION The use of Vibration analysis as part of maintaining rotational mechanical systems is well known. Diagnostics of mechanical systems using Vibration signatures have been researched in academic frameworks in connection with gear diagnostics, helicopter diagnostics, robots, ship Vibrations, tool-ware monitoring, and transportation. Another applica- tion is power plant monitoring systems, especially those used in nuclear plants, which frequently use Vibration sig- natures to identify worn parts requiring maintenance and other faults. In an aircraft engine, each machine (combination of parts applying energy to do work), such as fan, compressor, turbine, and gear box, has a unique and repeatable Vibration signature. Because the levels, profile, and features of these Vibration signatures correlate well between runs of the same engine, as well as between different engines of the same type, Vibration signatures can be a useful diagnostic tool. The high correlation between the levels, profile, and features of each run, for each machine mentioned above, can be seen in FIG. 1, which illustrates Vibration signatures from two runs for an exemplary engine, and in FIG. 2, which illus- trates the similarity in levels, profile, and features of vibra- tion signatures of different engines of the same type. FIG. 1 illustrates frequency domain analysis data from two runs on the same machine. As one of ordinary skill in the art will appreciate, the data collected during the runs is substantially identical. The levels and locations of peaks are almost the same between the two runs. FIG. 2 illustrates frequency domain analysis data from two runs on two engines of the same type. As may be appreciated, the collected data is closely correlated between the two runs. Generally, monitoring a mechanical system by analyzing Vibration signatures begins with collecting Vibration data at Various points in the system using Vibration sensors. The data is analyzed manually, electronically, or by a combina- tion of the two, to determine whether the data reflects normal or abnormal conditions of the mechanical system. The Vibrations represent the structural, dynamic, and aerody- namic characteristics of the observed components. In this manner, abnormalities, such as cracks, deformities, defec- tive parts, and deteriorated engine modules may be diag- nosed and the necessary maintenance may be performed. In addition to using Vibration signatures for diagnostics, recent attention has been given to the use of Vibration data for trend analysis, or prognostics, in mechanical systems. Trend analysis is generally concerned with identifying an abnormality at its incipient stage. Trend analysis is a valu- able tool, which enables one to proceed with corrective steps before an abnormality grows to a more costly, or even catastrophic, condition. Diagnostic and prognostic approaches to mechanical sys- tems using Vibration signatures have continually progressed. 10 15 20 25 30 35 40 45 50 55 60 65 2 For example, recent approaches include expanded automation, so as to significantly reduce dependence on a human operator. Moreover, while earlier approaches required shutting down operations in order to install a diagnostic apparatus, take measurements, and perform the necessary analysis, more recent systems have been designed to allow for online data collection. Finally, the tools for analyzing the data to reach a diagnosis or prognosis have become more sophisticated, and therefore more sensitive to abnormalities and trend data. However, current systems do not allow for concurrent data collection and data processing. Furthermore, the uti- lized Vibration analysis is usually limited to narrow band- width spectrum and to a single domain, such as the fre- quency domain. In other words, the Vibration signal is represented as a function over a set of frequencies. Typically the Fast Fourier Transform (“FFT”), is used to provide the representation of the Vibrational signature in the frequency domain. However, the FFT, because it is based on a single frame of data with a statistical error measured as 1/ (Number of frames), is statistically unreliable. Even recent applica- tions that have turned to the power spectral density (“PSD”) for Vibration analysis, because it provides higher reliability than the FFT, are generally limited to a single domain. Such FFTs are discussed in A. Mertins, Signal Analysisi Wavelets, Filter Banks, Time-Frequency Transforms and Applications (John Wiley & Sons, 1999), hereby incorpo- rated by reference as if fully set forth herein. Some applications calculate the PSD using Auto Regres- sive Moving Average (“ARMA”) modeling. ARMA model- ing is used to detect structural frequencies of the machinery as a rigid body and structural frequencies of its constituent parts. The vibrations spectrum/ spectrogram is estimated using ARMA model parameters. It is known that the spec- trum obtained with ARMA modeling emphasizes the struc- tural frequencies better than the FFT based spectra. A spectrum estimated using ARMA modeling is equivalent, with respect to the signal to noise ratio (“SNR”) of the result, to an average of 1000 frames in the traditional FFT-based PSD. ARMA modeling is also described in A. Mertins, Signal Analysis7Wavelets, Filter Banks, Time-Frequency Transforms and Applications. SUMMARY OF THE INVENTION A robust and efficient diagnosing and trending method and system is provided for use in the maintenance of a mechanical system, such as an aircraft, or power plant. The invention is based on a multi-domain, wide-band analysis of the Vibration patterns of various components of the mechani- cal system, which reflects the health of the corresponding components. The term Vib-RAY is used herein to refer to the multi-domain, wide-band Vibration analysis of the present invention. In accordance with the method of the present invention, sensors collect Vibrational data. The Vibration signatures of the monitored components are derived simultaneously in several domains: time, frequency, quefrency, time- frequency, order, amplitude, parameters, RPS-frequency (Rotations Per Second), and cycles. Because Vibration sig- nals in different domains, emphasize different faults, by correlating multiple signatures from various domains, the reliability of the diagnosis increases and the number of false alarms decreases. Every fault type of a monitored component is associated with at least one pointer, defining a frequency region of a Vibrational signature in a particular domain. At each pointer, US 7,027,953 B2 3 the current Vibrational pattern of the component, within the observed frequency region, are compared with a baseline pattern, using preferably up to nine mathematical operators referred to as “diagnostic indices.” The set of values pro- vided by each index when the pointer value is entered into the index is referred to herein as a vibration feature. The index is a function that provides a result by reference to a deviation from an expected “normal.” The features from the vibration analysis may be combined with features from gas path data of the mechanical system being monitored. Gas path parameters are physical param- eters that characterize operational conditions of the mechanical system. Examples of gas path data include fuel flow, pressure or temperature at various locations and engine stages, oil pressure, and shaft rotating speeds. The extracted features are then aggregated, quantized, and classified using a variety of artificial intelligence techniques, including neu- ral networks, support vectors machine, fuzzy adaptive reso- nance theory (“fuzzy-ART”), K-nearest neighbor, and expert systems, such as fuzzy logic and Bayesian networks. Thereafter, a hybrid artificial intelligence technique is used to diagnose and/or provide a prognosis for the monitored mechanical system. This decision process includes compar- ing the aggregated and quantized features to baseline fea- tures for particular pointers so as to determine whether the component is operating under normal or abnormal condi- tions. Because the system Vib-RAY is sensitive to signature changes and is focused on specific failure modes, it can distinguish between normal and abnormal states and can therefore diagnose abnormal patterns, as well as predict failures, by detecting problems at their incipient stages. The wide-band multi-domain analysis is effective in detecting cracks in blades, degraded bearings, engine compressor stall, damaged gearboxes, and improper assembly. With appropriate pointer identification and detection process, many other abnormalities can be detected, such as degraded gears and clogged nozzles. Furthermore, in accordance with the present invention, the vibration data may be measured in a non-invasive manner whereby data is collected by a sensor attached to the outer case of the engine, or turbine, or gearbox. Consequently, measurements may be taken during normal operation of the engine, helicopter, or other rotary mechani- cal system being monitored. The data can thus be advanta- geously sampled and analyzed online. Because the diagnosis and trending operations may be performed in either real time or oflline, Vib-RAY has general applicability for automatic diagnostics of mechanical systems and especially rotating machinery. Applications include engines, power-plants (e.g. aircraft, helicopters, marine, trains, ground vehicles, elec- trical power generation), drive-trains, gears and transmission, rotors, propellers, generators, and pumps. The present invention is particularly suitable to be imple- mented on-board an aircraft and in real time. Unlike prior art technology which require active off-line data mining, the present invention enables on-board, or real time, health diagnostics for rotating machinery. BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 illustrates a comparison of vibration signature data from different runs on the same engine; FIG. 2 illustrates vibration signature data from different engines of the same type; FIG. 3 illustrates a systems level architecture of an embodiment of the invention; 10 15 20 25 30 35 40 45 50 55 60 65 4 FIG. 4 illustrates further modules associated with the system shown in FIG. 3; FIG. 5 illustrates a diagnostic process in accordance with the present invention; FIG. 6 is a flowchart illustrating a portion of a vibration signal processing algorithm of the invention; FIG. 7 is a flowchart illustrating a portion of the vibration signal processing algorithm of FIG. 6, which is applicable to stationary parts; FIG. 8 is a flowchart illustrating vibration signal process- ing for non-stationary parts; and FIG. 9 illustrates analysis performed on features in accor- dance with the present invention. DETAILED DESCRIPTION Although the following description refers to aircraft and engines, as may be appreciated, the subject invention is applicable to other type of machines, including helicopters, marine vehicles, land vehicles, electrical power generation stations, trains, engines, gear boxes, drive trains, rotors, generators, and pumps. Analysis Domain Overview The prior art vibrational signal analysis for monitoring the health of mechanical systems have largely been confined to analysis in the frequency domain. For example, both the FFT and the PSD spectrograms described above are in the frequency domain. This means the spectrograms depict a power or other value at different measured frequencies of the machine. However, other domains may be useful for health maintenance. These domains include time, order, quefrency, time-frequency response (“TFR”), amplitude, parameters, rotations-per-second (“RPS”) frequency, RPS-order, and cycles. Indeed, each of these domains has particular utility for various aspects of system diagnostics or prognostics since each emphasizes a different source of vibration. Cur- rent applications generally do not analyze vibrations using more than a single domain. Consequently, other information from the vibration data that may shed light on the health of the mechanical system is lost. For example, the order domain is useful for analyzing the integrity of rotating parts because it eliminates the depen- dency of the vibrational spectrum or pattern on the varying rotating speed by concentrating the energy at a specific point. The order domain is derived by calculating the spectrum of the synchronized vibration time series with the rotation speed. After synchronization, the time domain is mapped to the cycle domain. For diagnostics, however, both synchronous and asyn- chronous data is necessary. The sampling rate is synchro- nized with the rotating speed of the monitored machine in order to convert from the frequency domain to the order domain. However, the detection of structural changes (e.g. cracks, blades tip degradation) requires asynchronous data because structural changes are manifested in the frequency domain usually by a shift in the natural frequency, which is provided by a PSD of the asynchronous data. Because existing applications are usually limited to either synchro- nous or asynchronous data, both the order and the frequency domains are not generally used by existing diagnostic sys- tems. Another domain useful for vibration analysis is the que- frency domain. The quefrency domain is also known as the Cepstrum domain and is the Inverse Fourier Transform (“IFT”) of the logarithm of the power spectrum, See Ency- clopedia of Vibration 2001 Vol. 1, page 216, hereby incor- porated by reference as if fully set forth herein. Cepstrum US 7,027,953 B2 5 enables the automatic detection of side-bands in the fre- quency or order domain, which appear due to modulation. Because the effects of modulation are known to represent defects in gears and bearings, Cepstrum is widely used to detect such defects. The use of the Time-Frequency Representation (“TFR”) is advantageous for the analysis and the representation of non-stationary signals (i.e., signals reflecting Vibrations measured when the rotational speed varies). When the rotating speed varies, the resultant change in operation conditions changes the Vibration pattern. The TFR describes the behavior of the spectrum as a function of time, which can be expressed as a function of the rotation speed to eliminate the rotation speed variability. The classic FFT-based spec- trogram is an example of a widely used TFR and is calcu- lated by first dividing the period into sequential short segments and then calculating a spectrum for each segment. The resulting spectrogram’s drawbacks include the assump- tion of partial stationarity in each time segment and a compromise between the time and frequency resolution. When the spectrogram is calculated over short time seg- ments the time resolution is improved, while the frequency resolution is degraded. Hence, the frequency resolution is inversely proportional to the time segment. The corollary is also true, when a large segment of time is used, the time resolution is degraded and the frequency resolution is improved. Improvements include the techues of overlap- ping and zero padding, which overcome the classic FFT spectrogram resolution problem. The overlapping technique consists, in part, of overlapping the time segments, which may be longer than the non-overlapping segments so as to provide improved frequency resolution while maintaining a better resolution in the time domain, since the time resolu- tion is equal to the resolution provided by the time segments prior to overlap. When each group of sampled points belong- ing to a segment (frame of time for spectrum calculation) are zero padded, the resulting frequency resolution is improved (depending on the number of zeros padded and the original time series length) without affecting the time resolution. Another example of a TFR is Wavelet transformation, which is primarily useful for acoustical analysis, as is known in the art. A WigneriVille Transformation is a third example of a TFR. This representation has the best joint resolution in time and frequency domains. However, the calculation results in a large matrix, requiring extensive processing. A fourth example of a TFR is the ARMA SpectrogramiTime Dependent model, which is adequate for random wide band signals of short duration. However, because most of the vibration signals measured on rotating machinery have harmonic (narrow band) components, the ARMA spectrogram is not well suited for such application. The above-mentioned TFR techniques can be used to represent the signal in the RPS-frequency domain or, after re-sampling, the RPS-order domain and statistical moments (in the time or RPS domains) including Root-Mean-Square (RMS), skewness, and kurtosis, which are respectively the 2’”, 3rd and 4th statistical moments of the probability density function of the signal. For vibrations signals the RMS represents the average energy of the signal. More specifically, the RMS generally represents the variance or the probability density function width. The skewness mea- sures the symmetry of the probability density function. The kurtosis represents the number and intensity of spikes in a signal. Each one of the above statistical moments can be calculated either in a specific frequency range or over the entire range. Overall figures can be calculated over all the measurement periods or over specific time slices, corre- sponding to specific rotation speeds. 10 15 20 25 30 35 40 45 50 55 60 65 6 Each of these domains provides a different benefit for use in vibration analysis. Different engine components have different base frequencies and effective correlation is per- formed at the third and higher harmonics. Therefore, wide band analysis is important for proper detection of abnor- malities. For example, the base frequencies of blade pass, bearings, and gear mesh, are 20 KHZ, 4 KHZ and 10 KHZ, respectively. To detect changes in these components, and to discriminate between pointers, it is important to analyze the vibrations over a large bandwidth. The typical bandwidth used for vibration analysis is from 0 KHZ up to about 275 KHZ. Process Overview The automatic trend diagnostic process is described below in detail, but as an overview, it is combined from three main stages. The first stage is data processing. This stage includes data evaluation, outlier’s elimination (eliminating clearly invalid data) and trend smoothing. Although the principles of data processing are the same for all data types, there are significant modifications adjusting the processing methods and parameters. For example, data that is collected by flight coupon (manual recordings of critical parameters) has many outliers that should be eliminated or augmented. The second stage is the Feature extraction, i.e. numerical representation of the monitored parameters characteristics. The features can be parameter deviations from the initial- ization point (snapshot), or shift of each parameter over a number of cycles. The basic features in current use are: snapshot, short-term shifts, long-term shifts, and varying- terrn shifts. It should be noted that different features provide different information about the engine. For example: snap- shot and short-term shifts provide information on abrupt changes, as broken valves and open bleeds. Long-term shift are more appropriate for detection of slow deterioration of engines. The third stage is classification. Each of the features is classified by several diagnostic methods. The results of the various diagnostic methods are aggregated so as to increase detection confidence. As may be appreciated, there is sub- stantial interaction between components associated with each of the three stages, such as, for example, by requesting collection of additional data or requesting processing of data in domains and pointer relevant to a detected condition. Furthermore, some of the stages are combined as a single stage in various embodiments of the invention. System Architecture Referring to FIG. 3, an overall system architecture of an embodiment of the present invention is shown. FIG. 4 illustrates a lower level view of the same architecture. The subject mechanical system of this illustration is the engine 101 of aircraft 100. Sensors 110 are used to acquire engine and aircraft data. Vibration sensors 111 measure vibration data and path sensors 112 measure gas path parameters, from various selected components of the engine 101. Gas path parameters are physical parameters that characterize opera- tional conditions of the engine. For example, fuel flow, pressure/temperature at various engine stages, oil pressure, and shaft rotating speeds are gas path parameters. Gas path data types can be divided into two major categories: repeatable and non-repeatable/transient data. Repeatable data is recorded during predetermined flight regimes. This data can be measured during the following flight phases: run-up, takeoff, cruise, top of climb, and end of descent. Non-repeatable/transient data collection is trig- gered by events such as over limits alerts, as discussed below. This type of data allows the system to analyze the engine operation during the event and generate new features US 7,027,953 B2 7 for automatic diagnostics. The present discussion will only address repeatable data. The Vibration data and gas path parameter data are transferred over a communications bus 115, to the health and maintenance on board unit 200 for further processing. The on board unit 200 includes an aircraft communications interface through which it is coupled to the bus 115. Employing the communication bus 115, the on board unit 200 receives additional information from the engine 101, such as pressure, altitude, ambient temperature and engine control status, which is not measured by the sensors 110. There are three major resources for data collection. The first resource is flight coupons, which include manual recordings of critical parameters. These parameters are preferably collected only during cruise. These are forms manually filled by the pilot during flight, which include data related to gas path parameters, including speed, altitude, ambient temperature, EGT, pressures, and fuel flow). Flight coupons are generally used in aircraft that do not have ACARS. The reliability of this data is generally lower than other data due to the manual nature of data entry. A second resource is ACARS telegrams, which include automatically collected data. Simple algorithms are used to determine when the aircraft is in a certain predefined flight regime. When a certain flight regime is attained, data is recorded and evaluated for quality. Such data is preferably collected only in a limited number of flight regimes. A third resource is sensor Vibration data, which is collected automatically. This resource allows for recording non-repeatable data, when triggered by predefined events (e.g., over limits, Full Authority Digital Engine ControliFADEC alerts). A cor- responding flight regime detection algorithm allows for detecting additional flight regimes, e.g., top of climb or end of descent. The engine control 116 is a general nomenclature for any electronics box, originally installed on the engine by the OEM (e.g., Electronic Engine ControliEEC, FADEC), which is generally used to control the engine operation and its accessories and provide for high level remote operation of the engine by the aircraft systems (such as the flight control computer, the automatic thrust management, and the maintenance computer). Additional parameters required by the engine control 116 for proper operation, are provided by the Air Data Computer (ADC), which supplies environmen- tal conditions data, used by the Engine Control for compen- sation and adjustments. The engine control 116 also provides measured and processed engine parameters data to the aircraft systems (e.g., Display Electronic UnitiDEU, Cen- tral Fault Display Interface Unit%FDIU, and Digital Flight Data RecorderiDFDR). The on board unit 200 also receives flight data from the aircraft, such as flight type, speed, and environmental con- ditions of the flight, as shown by block 113. The charge amplifier 210 is the first stage of the vibration sensors 111 interface. The charge amplifier 210 is used to filter and amplify the electric charge generated by the vibration sen- sors 111. The charge amplifier 210 converts the electric charge signal into a voltage signal for further sampling and processing by the vibration and data analysis (“VADA”) module 230. Although the on board unit 200 is a non-essential system for proper operation of the aircraft, it is nonetheless envi- ronmentally protected as though it was an essential system because it interfaces with critical systems such as the engine control 116. The buffers module 220 is designed to provide sufficient protection to the on board unit 200 against loading on critical signals, induced lightning effect, electromagnetic 10 15 20 25 30 35 40 45 50 55 60 65 8 interference (“EMI”), and High Intensity Radiated Fields (HIRF). The buffers module 220 also isolates the gas path sensors 112 from the on board unit 200. Gas path and flight data are processed by the processing unit 240. The VADA module 230 samples and analyzes the vibration sensor signals. The primary purposes of the on board unit 200 is to (i) monitor the behavior, integrity and performance of the engine drive train and other engine machineries; (ii) detect and document engine anomalies; and (iii) draw conclusion on their status. In a preferred embodi- ment of the present invention, the engine and aircraft con- ditions are monitored from Power ON to Power OFF. As described more fully below, diagnostic algorithms embed- ded in the on board unit 200 detect engine anomalies or other abnormal situations. When the monitored data exceeds pre- defined limits, such as when the engine temperature exceeds a threshold level or other deviations in flight data, as compared with prior flights, the onboard unit 200 records certain identification information and the history of the monitored parameters onto data storage 250. The identifi- cation information preferably includes an advisory fault code, the date and time of the event, and the aircraft and the engine serial number. Various parameters are also recorded when there is a deviation that exceeds a limit. The objective being to provide as much information as possible about the conditions and possible causes of the deviation. In general, all applicable monitored parameters are recorded for a predetermined time frame. All applicable monitored param- eters are configurable to the specific engine and aircraft types as well as to the particular installation. In one embodiment, the monitored parameters’ history information includes all parameters, other than vibration data, that were monitored during the time period beginning with 10 seconds before the abnormal event was detected and ending with the disappearance of the abnormal event, up to a maximum of 3 minutes. Additionally, twenty seconds of vibration data is recorded during a predetermined timeframe. For example, the time can be set to 10 seconds prior and 10 seconds after the event. The on board unit 200 also records data at certain pre- defined normal aircraft operational states or flight regimes. These states preferably include takeoff, climb, cruise, and landing. During these states, the on board unit 200 records identification information and the history of the monitored parameters. The identification information preferably includes a trend identification code, the date and time of the event, the aircraft identification and the engine serial num- ber. The monitored parameters’ history information includes all parameters, including vibration data, from 10 seconds prior to the event until 10 seconds after the event. Finally, all monitored parameters are recorded during engine start and engine shut down. The output of the on board unit 200, as described below, may be communicated to an off-board, or ground, station 400 for further processing. Such communication may be in real time or at a pre-defined schedule using, for example, GSM wireless communication. Alternatively, the on board module output may be stored in a data storage unit 300 from which the data can be transferred to the ground station 400 at a later time. Based on pre-defined algorithms, the on board unit 200 computes and records Life Usage Indicators (“LUI”). The LUI are recorded as part of the abnormal event parameters, which provide the cumulative engine operating data asso- ciated with the event. During normal operating conditions, the LUI are stored by the T-HUMS computer, in its local memory, as opposed to the Mass Storage memory where the US 7,027,953 B2 9 abnormal events are recorded. LUI are accumulated param- eters retrieved on Power ON, updated during the Power ON cycle, and are recorded upon Engine OFF or Power OFF. LUI includes: low cycle fatigue (LCF), high cycle fatigue (HCF), engine operating time (EOT), number of starts (NOS), engine flight time (EFT), time above dwell temperature, accumulated reverse thrust time, Max Exhaust Gas Temperature (EGT) at start, max fuel request at start, max N2 (free compressor-turbine shaft rotational speed, indicated in Revolutions Per Minutes) at start, oil quantity reduction at start, start stage time, and Shut Down time. In addition to accumulated data, the data storage 250 also contains the knowledge base of normal and abnormal fea- tures against which current and trend features are compared to determine an appropriate diagnosis and prognosis of the mechanical system, as described in more detail below. As will be readily apparent to one skilled in the art, the system architecture as described is flexible and configurable to support various engine types and aircraft platforms. Data Collection Referring to the block diagram of FIG. 5, the diagnostics steps of the present invention are illustrated. Blocks 410 through 413 represent the initialization of the diagnostic method of the present invention, or measurement setup (410). The initialization includes connecting vibration and gas path sensors at appropriate locations (413) and identi- fying the parameters to be captured (411). Additionally, flight regime data (415), such as, speed, length, and envi- ronmental conditions are input into the on-board database 250, illustrated by the arrow leading from block 473 to block 250. The database 250 also tracks (i) identifying indicia for the monitored aircraft, engine, and gear box (block 420), (ii) the maintenance history of the aircraft (block 430), and (iii) the knowledge-base of normal and defective vibration signa- tures in the various domains of interest (block 440). Vrbration data is sampled from each vibration sensor at an approximate rate of 50 KHZ to 100 KHZ, at a preferred bandwidth of 20 KHZ to 50 KHZ. Employing a segment length of 20 seconds and four sensors, each segment results in four million to eight million words, or samples of vibra- tion data. As vibration data is collected, vibration signatures are created in a number of domains (block 450). Features are extracted in block 460 and are compared with the features of known signatures by artificial intelligence techniques, as described below (block 470). When a novel pattern of features is detected, a new diagnostic cycle is triggered, beginning with the Failure Mode Analysis (block 480). The Failure Mode Analysis is a preliminary phase that includes exploration of the specific failure, understanding its mechanism, determining system signature impact, and identifying the affected parameters, including root causes. The result of the failure mode analysis determines the Measurement Setup to follow (decision upon measured parameters, measurement location, bandwidth etc.) (block 410) and the appropriate signatures of Signal Processing (450 block). Specifically, the signature types are advantageously selected depending on the known failure modes and their mechanism so as to focus the observation on the domains and pointers relevant to the expected phenomena, or failure mechanism. Subsequently, new fea- tures are extracted and the decision process is adapted. Database 250 is updated with the new feature data and the decision process intelligence is retrained, as shown by block 490. For Expert Systems, the retraining is preferably achieved by expanding the rule base. 10 15 20 25 30 35 40 45 50 55 60 65 10 Signature Generation The steps for generating signature data in one embodi- ment of the invention will now be discussed by reference to FIGS. 6, 7 and 8. Appendix A illustrates the specific algo- rithms associated with each function for providing a signature, as described below. FIG. 6 illustrates the common processing stage for sig- nature generation for all parts. Not shown in FIG. 6, but prior to signature generation, the vibration data is preprocessed to determine quality and validity. This is determined by refer- ence to statistical considerations. For example, the system validates the data collected from the sensors to ensure that the data range is reasonable or realistic by reference to this statistical data. In addition, preprocessing generates representative values for each parameter. The input parameters (features) used for automatic diagnostics are selected to emphasize character- istics that are correlated to specific failure modes. Data processed from different engine operating regimes enables us to emphasize deterioration phenomena that are more explicit in certain flight regimes. When more than one flight regime reflects a specific failure mechanism, data redun- dancy is used to increase the diagnostics reliability. The raw data is preferably calibrated from 16 bit integers to voltage and onto physical units, depending on the gain for each specific channel, which is usually in G units for vibrations (601). Rotating speed analysis is performed, entailing frequency determination using periods between zero crossings, and smoothing using a moving average (611). Zero crossing and smoothing are described in Sec- tions 1.1 and 1.2 of Appendix A, respectively. Statistical moments as a function of time (DC, RMS, Skewness, Kurtosis) are calculated, with a preferred resolution of 0.1 sec (603, 605, 607, 609). Next, the data is classified as stationary or non-stationary using the analyzed rotating speeds and the statistical moments of the vibrations (614). Data is preferably classi- fied as steady state, stationary data, when it is valid data with changes in the RPM and vibrations RMS in a range of 110% of the mean value. Data preferably is classified as fast acceleration, non-stationary data, when it is valid data with monotonic increasing RPM of more than 30% in less than 5 sec. Data is preferably classified as slow acceleration, non- stationary data, when it is valid data with monotonic increas- ing RPM of more than 30% in more than 5 sec. Data is preferably classified as fast deceleration, non-stationary data, when it is valid data with monotonic decreasing RPM of more than 30% in less than 5 sec. Data is preferably classified as slow deceleration, non-stationary data, when it is valid data with monotonic decreasing RPM of more than 30% in more than 5 sec. FIG. 7 illustrates signature generation for stationary parts. The system of the present invention calculates signatures in the PSD (617), which uses at least 50 frames for each PSD, resulting in 86% reliability (Such PSD calculation is described in Mertins. A., “Signal Analysis wavelets, Filter Banks, Time-Frequency Transforms and Applications”, John Wiley & Sons, 1999, Chapter 5). The PSD is used mainly for detecting structural changes in the observed system. As mentioned above, corresponding features are extracted (not shown) from this signature data. The PSD function is defined with more detail in Section 1.7 of Appendix A. The system also calculates Cepstrum signatures for sta- tionary parts (Such calculations are described in Braun. S. G, Ewins. D. J., Rao. S. S, “Encyclopedia of Vibration”, Academic Press. 2002. pp 2167227, 7477748). Statistical Moments (of certain bandwidth) are calculated from the data (619). The moments preferably include DC, RMS, skewness and kurtosis. US 7,027,953 B2 11 The Vibration signals are also re-sampled according to each of the relevant rotation speeds for defined failure modes (615) (described in Lyon. R. H., “Machinery Noise and Diagnostics”, RH Lyon Corp 1998, Chapter 6). The re-sampling operation (615), as well as dependent operations, are repeated for all rotation speeds. The re-sampling function is defined in more detail in Section 1.5 of Appendix A. Using the re-sampled Vibration signals (615), the system of the present invention preferably calculates signatures three different ways. The Cepstrum signature (631) is cal- culated from the Orders-PSD (625), described in Lyon. R. H., “Machinery Noise and Diagnostics”, Chap. 6 (RH Lyon Corp. 1998), hereby incorporated by reference as fully set forth herein. Features will be extracted from this signature, as described in more detail, below. The Cepstrum function is defined in more detail in Section 1.9 of Appendix A. The Cepstrum-Orders-Phase signature (639) is derived by first calculating the Phase mean with filter (621), described in Lyon. R. H., “Machinery Noise and Diagnostics” Chap. 7 (RH Lyon Corp 1998), hereby incorporated by reference as if fully set forth herein and then the Orders-Phase average is calculated by employing the PSD data (635). Again, features will be extracted from this signature. The Phase Mean function is defined with more detail in Section 1.6 of Appendix A. In addition, the Statistical Moments-Phase average may be calculated from the Phase Mean (621) as a function of the rotation speed that is synchronized with each observed part (637). Lastly, another Cepstrum signature is derived by calcu- lating the Envelope with filter from the resampled data (623), described in Lyon. R. H., “Machinery Noise and Diagnostics” Chap. 6.3 (RH Lyon Corp 1998). The Enve- lope function is defined in more detail in Section 1.8 of Appendix A. In addition, the Orders of the Envelope signa- ture may also be derived, (627) and corresponding features are then extracted. Finally, Statistical Moments, including RMS, Skewness, and Kurtosis, are calculated from the resampled data (629). The statistical Moments function is defined in more detail in Section 1.11 of Appendix A. FIG. 8 illustrates signature generation for non-stationary parts. For Non-stationary parts, where the rotation frequency is not constant, a system of the present invention preferably calculates signatures by employing several domains. The Spectrogram is calculated to provide a time-frequency rep- resentation of the data (641) and is described in Braun. S. G, Ewins. D. J., Rao. S. S, “Encyclopedia of Vibration” (Academic Press. 2002), pp. 216727, 7477748, hereby incorporated by reference as of fully set forth herein. The Spectogram function is also defined in more detail in Section 1.13 of Appendix A. The WigneriVille is calculated to provide a second time-frequency representation of the data (647), described in Braun. S. G, Ewins. D. J., Rao. S. S, “Encyclopedia of Vibration” (Academic Press. 2002), pp. 216727, 7477748, defined with more detail in Section 1.15 of Appendix A. The Background Spectrogram of the data, defined with more detail in Section 1.13 of Appendix A, is calculated in 645, followed by extraction of corresponding features. The data is also re-sampled according to each of the relevant rotation speeds for observed parts (643) and using the resampled data the spectrogram by Orders, providing a time-orders representation, is calculated (649). Features from all of these signatures are also extracted. The system of the present invention next preferably normalizes the signatures to a common denominator to enable comparison between signatures. For this purpose, a 10 15 20 25 30 35 40 45 50 55 60 65 12 model is built, which correlates vibration energy to flight conditions where dynamic pressure is the main factor. This normalization is accomplished by resealing each signature according to this model. This generates about 2 Mega words of signature information. The system then generates the baseline signatures by taking the weighted average of sig- natures acquired in equivalent operating conditions of a specific system, engine, and gearbox, which are known to be in good condition. Feature Extraction In accordance with the present invention and as noted above, features have to be extracted from the various sig- natures that are calculated. Vibration features are used for the diagnostic and prognostic functions of the present inven- tion. The vibration features are diagnostic index values for a given pointer, in a given domain, for a given vibration signature. The diagnostic indices are mathematical functions which compare acquired signature data to reference, or baseline, signature data by reference to pointer information associated with the desired features. Pointers are calculated for every type of system, engine and gearbox, using known tech- niques. Such techniques are described in Braun, S. G, Ewins, D. J., and Rao, S. S, “Encyclopedia of Vibration” (Academic Press. 2002), and in Braun, S., “Mechanical Signature Analysis, theory and applications” (Academic Press 1986), both of which are hereby incorporated by reference as if fully set forth herein. The pointers vary by the mechanical makeup, assembly and physical dimensions of the compo- nents being monitored. On average, each aircraft system is associated with about 1007200 pointers. The number of relevant harmonics is defined for each identified pointer. The exact number of harmonics for each pointer depends on the pointer’s base frequency and on the physical phenomena it is associated with. For example, pointers related to rotating motion will appear with many harmonics bandwith ] d d‘ th t‘ — [ epen mg on e ra 10 pointer base frequency while pointers related to natural frequencies of a component will manifest without higher harmonics. Initially, each pointer is preferably calculated in the frequency domain and then converted for reference in other domains, as required. The type of conversion will depend on the domain to which it is converted and the fault character- istics for that pointer. For example, gear-shaft eccentricity manifests as a frequency modulation around the gear-mesh frequency. The frequency modulation is based on the inlet and outlet shaft rotating speeds. In the Cep strum domain, the pointers representing the frequency modulation are the inverse (1/x) of the respective shaft rotating speeds. Accordingly, the system employs the shaft rotating speed to calculate the relevant pointers. Once the pointers are calculated, up to nine diagnostic indices are preferably associated with each pointer. Although a variety of diagnostic indices may be used, in one embodiment a system of the invention employs: an arith- metic (“Amn”), geometric (“Gmn”), harmonic (“Hmn”), matched-filter RMS (“Mfrms”), RMS of spectral difference (“Rdo”), Sum-of-Squares (“So”), energy-ration, entropy, and frequency-shift indices. Each index is an operation on the current signature and a reference signature of the observed system. The reference signature is preferably a signature of the observed system for a known condition relating to system health. US 7,027,953 B2 13 The Arithmetic index defines the difference in decibels between the energy of the current signature and the reference signature. The Geometric index is the ratio between the geometric mean of the current signature and the reference signature. The Harmonic index is the ratio between the harmonic mean of the current signature and the reference signature. The harmonic mean is well known in the art, either as an arithmetic or as a geometric mean. The Matched Filter Root Mean Square (“Mfrms”) is the root mean square of the ratios between the two signatures. The RMS of spectral difference (Rdo) index is the root mean square of the differences between the two signatures in dB. The So index is the sum of squares of differences. Further details of these indices are provided in Appendix A. The energy-ratio index is the difference between the energy ratio of two adjacent and equal portions of the current spectrum and the similar regions of the baseline spectrum. This ratio emphasizes and detects shifts of frequency point- ers and it is mainly used for detection of natural frequency shifts, which manifest as structural changes in the observed system. The entropy index is the relative entropy of the actual signature as compared with the respective base line signa- ture. The entropy index measures the morphology of the signature around a specific pointer. The relative entropy measures the changes in the pattern morphology. The frequency shift index calculates the difference between the centers of mass of the actual signature and the respective baseline. The center of mass of the signature is calculated in a predefined region of the signature (around a pointer), representing the frequency (for spectra) where most of the energy is concentrated. The difference between the centers of mass of two signatures quantifies the fre- quency shift. Each index identifies a different type of change in the signature pattern in a region around a given pointer. The diagnostic process of the present invention employs the diagnostic indices to detect shifting, energy increases/ decreases, sharpness, or spread of the peaks of the signatures, or any combination thereof. The use of energy-ratio, entropy, and frequency-shift indices for vibration analysis has not been taught prior to the present invention. The other diagnostic indices mentioned above, have been used to compare parametric models (ARMA). See Braun, S. G, Ewins, D. J., and Rao, S. S, “Encyclopedia of Vibration” (Academic Press 2002) and Flandrin, P., “Time-Frequency/Time-scale analysis” (Academic Press 1999), both of which are hereby incorpo- rated by reference as of fully set forth herein. The calculation of the diagnostic indices leads to a 3-dimensional matrix of vibration features. For the exem- plary aircraft engine, there are approximately 500 pointers represented on the x-axis. The y axis represents the diag- nostic indices. For the exemplary aircraft engine there are an average of 9 indices for each pointer. The Z axis represents the domain in which a signature is shown. For the exemplary aircraft engine, there are about 5 domains. This combination results in approximately 22,500 features in the exemplary aircraft engine. After providing the features for the observed engine, the system generally proceeds to aggregate the features by employing a weighted averaging formula for the normalized features that are relevant to each particular fault. This includes features extracted from harmonics and sidebands of a primary feature or pointer, as well as features from different domains. Each feature in the aggregation is 10 15 20 25 30 35 40 45 50 55 60 65 14 assigned a weight which affects how the feature influences the overall determination arrived at by the aggregation operation. The main purpose of this formula is to correlate the pattern changes of all the harmonics over the signatures in all domains, which are applicable to a specific failure mode. For example, for a gearbox failure mode, features are aggregated from the first harmonic of the meshing frequency and corresponding sidebands, the second, third, and fourth harmonics of the meshing frequency and corresponding sidebands, and the first three harmonics of the shafts in the cepstrum domain. Furthermore, aggregation helps reduce the amount of data provided to the analysis components of the system. Prior to aggregating the features, however, the indices for each pointer are collected and normalized. Once normalized, the features representing the harmonics of each pointer are aggregated. Finally, features of the same pointer (of a particular failure mode) are aggregated across the different domains. Baseline signatures are derived prior to operation of the system so as to allow for comparison of acquired signatures to baseline signatures by employing the various indices. For purposes of this description, a new engine is an engine that operates within the acceptable operational range of a normal engine, and has been recently installed on an airplane, with raw data that varies as a function of environmental and flight conditions. The system of the present invention employs a baseline algorithm to provide a common baseline to all engines of the same type, i.e., eliminate installation and unique engine variation from nominal operation. This algo- rithm determines the transformation function, which brings each engine to vary around some nominal point. The base- line algorithm generates a compensation mechanism, which, with given flight and environmental input such as mach number, engine pressure ratio, rotation speed, air temperature, air pressure, and the like, is used to generate the nominal average engine parameters that are typical of a “new engine.” The baseline algorithm begins with an acquisition step where data is collected during the first operation period of the engine. A validity and range checking is performed to eliminate all invalid data while retaining valid data snap- shots. Thereafter, the new engine data is added to a baseline calculation database. When using a neural network to calculate the baseline, a different neural network is trained for each parameter trans- formation to a baseline feature. Thus, a baseline feature is calculated for each flight regime parameter so as to allow the system to later compare features from acquried data to a corresponding baseline associated with the same parameter and flight regime. The network inputs are the relevant environmental and flight data and the target is the appropri- ate parameter. Exemplary targets are N1 for the rotation speed-low pressure, N2 for the rotation speed-high pressure, EGT for the engine temperature, and FF for fuel flow. In one embodiment, the neural network that is used to create the baseline feature is a multi-layer feed-forward neural net- work. However, the architecture of the neural network may vary in the transfer function, number of cells in the hidden layers, and number of inputs. A general discussion of such networks is provided in Bishop. C. M., “Neural Networks for pattern Recognition” (Oxford University Press NY. 1999) pp 310. and Haykin, S. “Neural Networks” (Prentice Hall 1999), both of which are hereby incorporated by reference as if fully set forth herein. Similarly, when baseline feature calculation is performed using polynomial fitting, a different polynomial is fitted for US 7,027,953 B2 15 each parameter. The fitting criterion is preferably the least mean square (“LMS”) error with respect to the measured (actual) parameters from the “baseline calculation data- base.” The order of the polynomial and the input parameters used are the most relevant to the specific parameter. For example, for ANl the input parameter can be EPR and MN and the order of the polynomial can be 3. The polynomial order is preferably high enough to accurately map the normalization function but not so high as to cause an “over fitting” problem. Input parameters are the parameters that reflect environmental influences, which should be elimi- nated in order to obtain an accurate comparison. In some cases, the environmental influences include the airplane’s gross weight, and component temperatures. As in the poly- nomial order, the system finds the minimum set that provides an accurate normalization. At this time, the system has available a baseline feature, which is based on the nominal values for the new engine. A shifts transform is applied to shift the engine baseline of the “nominal” value to that of a specific engine. The differential between engines results from differing position and instal- lation on the airplane, and variations in manufacturing. The shift value is derived by calculating average differences between the actual values and the appropriate baselines. This shift is calculated based on the first 20 to 50 flights of an engine on a specific wing position. The baselines are pref- erably recalculated every time an engine is dismounted and remounted on a wing, because a different installation dif- ference will appear, requiring the calculation of a new shift. Additionally, the baseline is recalculated for any changes of the engine operation line due to maintenance. Although Gas-Path features are not required, nor are they applicable to every implementation of the present invention, where the invention is applied to aircraft, helicopter or any other complex mechanism, those features are preferably included, particularly for engines since they improve the diagnostic reliability by allowing for correlation between more types of features. As with the vibration data, features are extracted from the gas path data before any diagnostic or prognostic decisions are made. The gas path feature extrac- tion strategy involves several steps for all parameters from a single flight regime. These steps include checking data validity, estimating data quality, data transformations, cal- culating corrected parameters, calculating a baseline, calcu- lating deviation from that baseline, and calculating all types of features (including trending, gradients, and shifts). A quality check of the gas path data is performed for each flight, including each flight regime, such as takeoff, landing, and cruising, and for each feature extracted based on the gas path data. The data quality is preferably calculated after the data is acquired, but before the data is transformed into a useful feature. This quality check involves several steps. First, each parameter is examined to determine whether it is within its acceptable range, as determined by an engine expert or predetermined ranges. Next, the statistical moments of the input data are compared to ranges that have been determined by statistical checking of the historical data. As is known, statistical moments are used to measure the Probability Distribution Function (PDF) by an ordered integration of the PDF function. Since each parameter has different ranges for different flight phases, for example the altitude range at takeoff is different than in cruise, the data is also logically cross-checked with known flight parameters. Finally, the stability of the parameters during the measurements period is also verified to be within a valid range. Data validation is performed by ensuring that (i) each parameter is within its valid range allowed by the measure- 10 15 20 25 30 35 40 45 50 55 60 65 16 ment configuration; and (ii) that the flight condition is stable during the measurements period. Prior to deriving the features, the system preferably adjusts the raw data by the Laws of Thermodynamics so that the data follows standard environmental conditions. For example, the system corrects the flight parameters to fit to standard temperature (15 degrees Celsius) and standard altitude (sea level). The applicable transformation formulae for providing such adjustments are known in the art. Generally normalization is discussed in chapter 8 “Pre-processing and feature extraction”, Bishop, C. M, “Neural Networks for pattern Recognition” (Oxford University Press, NY. 1999), hereby incorporated by reference as if fully set forth herein. lnvalid gas path parameters are not considered in the decision process, or in the trending process. Once the gas path parameters are validated, they are transformed into the more meaningful features. For example, two pressure parameters, pressure 1 and pressure 2 are transformed into a pressure ratio Pressure 1/Pressure 2 which may indicate a pressure ratio variance beyond a normal level. Next, a baseline feature is derived for each gas path parameter. The baseline feature consists of a generic com- ponent based on engine behavior according to each engine type. The baseline is preferably calculated with either a neural network, or through polynomial fitting. By consider- ing accuracy, data type, data amount, and applicability, one skilled in the art may determine which technique is suitable for a given baseline feature. Another feature employed by a system in accordance with the present invention is the deviation from the baseline. This is preferably a separate value, which is derived for each relevant parameter. The deviation is preferably calculated on all relevant parameters from different flight regimes such as Takeoff, Cruise, and Taxi. Furthermore, in some cases the deviation is calculated for transformed parameters. The term “transformed parameter” refers to a combination of more than one parameter (e.g., ratio of two parameters such as pressure ratio). The flight regimes are characterized by repeatable and stable flight conditions, which facilitates comparison between different engines as well as follow up on the same engine operating under similar conditions. The deviation feature represents a change of a specific parameter in a specific flight regime from the nominal expected value, or baseline. The deviation is calculated by one of several methods. One method includes calculating the difference between the corrected measured parameter and the calculated baseline. This provides a percent deviation from the baseline. Engine operation lines set by the manufacturer define the normal ranges of operation. For example, the “Surge Line,” which can be derived from the compressor operation map provides temperature maps that combine the influences of the engine load environment conditions and engine temperature. The deviation of the engine from its operation line indicates the condition of the engine and its safety margin. Another example is the “Compression Ratio,” which defines the ratio between the compressor’s intake and exit pressure. The compressor map is a graph of the compression ratios relative to the engine thrust at constant rotation speeds. The com- pressor surge line is marked on the graph. The distance between the actual working point of the engine to the “Surge Line” provides the safety margin of the compressor. The “TAKEOFF Engine Temperature Margin” is another operating indicator that can be used to indicate a deviation. The engine temperature is tracked during TAKEOFF, where maximum power and load operation are present. The maxi- mum temperature value is compared with a maximum US 7,027,953 B2 17 allowed temperature provided by the manufacturer. The current EGT value preferably depends on flight conditions. The difference between the operation point and the allowed temperature indicates the ability of the engine to operate under the desired load. An engine can be divided into 4 modules, Low Pressure Compressor (LPC), High Pressure Compressor (HPC), Low Pressure Turbine (LPT), and High Pressure Turbine (HPT). Each module’s efficiency is calculated using the intake and exit pressure ratio (corrected for the specific environmental conditions) and engine parameters such as fuel-to-air ratio. As may be appreciated different module divisions are used for calculating efliciencies with different engine models and available parameters. Next, the system extracts diagnostic features from the treated parameters (ANl, AN2, AFF, AEGT, EGTiMARGlN). Several different diagnostic fea- tures can be extracted from the same set of parameters. For example: snapshot features represent the instantaneous condition, while short and long term trend features represent time dependant deviations in the machine condition. Trend features represent the deviation tendency of the engine parameters over time (cycles). The deviation ten- dency is used to diagnose and predict development of engine faults. The constant trend shift method uses constant time (cycles) to evaluate the trend shift for each parameter within the cycle range. The cycle ranges preferably include short- term shifts of about 10 cycles and long-term shifts of about 90 cycles. In one embodiment, the system also extracts varying term features. These features are values extracted by an automatic algorithm, identifying break points in the trend curve. The Curve breaking point algorithm detects the point (cycle) where the curve starts to change. For this purpose, the system employs several methods: (1) deviation from a predefined linear fitting, (2) polynomial fitting and analysis of the fitted polynomial local minimum or maximum, and (3) estimation of the correlation matrix for the trend begin- ning and finding the point where this correlation is no longer valid. The Varying Feature algorithm examines the shift of each parameter from a common breaking point, and the number of cycles from the breaking point. Feature Classification In a typical aircraft engine, the aggregation of the features leads to approximately 200 features. As will be appreciated by those skilled in the art, for other components, aggregation may result in more or less than 200 features. As is illustrated in FIG. 5, once features are extracted (block 460), the trending process (block 461) is initiated. As used herein, trending refers to the construction of a trend curve. Trend analysis refers to the methodology, which is used to analyze the curve. The trend analysis, shown in block 462, involves calculations of the gradients of the trend curve and the corresponding shifts of the trended parameter. These gradi- ents and shifts become new features. If the trending process or the calculation of the aggregated final features is per- formed on-board, the data is preferrably transmitted to a ground station where it is analyzed. As described below with reference to the diagnostic process, the diagnostic process is repeated on the trend features to predict data. Decisions are made by reference to predicted data from the prognosis function of the present invention. Another operation of the diagnostic sequence is verifica- tion of over limits. A verified over limit results in an alert. The system of the present invention preferably provides the following data to assist the engine expert in analyzing the over limit event and determine its criticality: a record of all parameters before, during, and after the over limit event; 10 15 20 25 30 35 40 45 50 55 60 65 18 relevant diagnostics history of the engine; and supporting information such as the engine maintenance schedule. The diagnostic process also detects aircraft sensor failures, which are characterized by simultaneous trend shifts of a specific parameter in all of the aircraft’s engines. Snapshot diagnostics uses a Fuzzy ART (Adaptive Reso- nance Theory) neural network. This is preferably an unsu- pervised learning classifier, which partitions the parameters’ multidimensional space into groups. Any unknown fault that is reflected in the snapshot data will be classified as Novelty until the expert identifies the fault. This classifier continu- ously improves as more data and feedback are accumulated. The principle of the trend diagnostics is to detect relative changes of the parameters A in respect to previous measure- ments. The automatic diagnostics sequence is combined from features extracting, multi classification methods and finally decision processes that determine the engine condi- tion and the confidence level of the diagnostics. The novelty detection method is preferably used to aflirm the classifier’s results. Before extracting trend features, outlier elimination and smoothing operation are preferably performed. Outlier elimination omits data that is clearly a measurement error. The smoothing procedure uses statistical methods such as moving average, to emphasize the trend behavior. Fitting a linear curve to the specific parameter trend is used for trend feature extraction. The vertical distance between the begin- ning and the end of the fitted-line is the feature (Shift). The average fitting error is used to determine the confidence level of the classifications that are based on that feature. There are preferably three main feature types provided by the system. A short term feature is one where a shift is developed during a small number of flight cycles (usually between 10 to 30). This feature is used to detect sudden changes. Examples of failures which manifest in such abrupt changes include broken parts, stuck valves, and open bleeds. A long term feature is one where a shift is calculated over a large number of flight cycles (80 to 120). This feature type is generally used to detect deterioration in engine perfor- mance. A varying term feature is one where a shift of trend parameters from an automatically detected break point is observed. This feature type is used to strengthen both previously discussed feature diagnostics. The trend break point is detected by reference to statistical variation in the synchronized trend parameters. Each one of these features is preferably diagnosed by several classification methods. Each one of the selected classification methods has it own advantage, and the overall method combination is selected to support other decisions. The use of more than one feature and diagnostic method improves the sensitivity and the reliability of the diagnostics decision. The following discussion briefly explains the prin- ciples and benefits of each diagnostics method. Expert system methods facilitate embedding expert knowledge into the diagnostic process of the extracted features. The nearest neighbor method provides a variation of the neural network classification method that is appro- priate for use with expert knowledge. The diagnostics sequence is combined from the following steps: Normalization of the data to a common scale [(Ll] for comparison purposes, Expert knowledge, i.e., failure modes definitions, which is mapped into specific states in the features space, The distance (in the feature space) between the input data and the failure states is calculated, The class with the shortest distance from the given input state is selected. Fuzzy Logic Expert knowledge is embedded in the deter- mination of Fuzzy Sets (FS) and in the Fuzzy Rules (FR). ln US 7,027,953 B2 19 this method, normal engine operation is considered as one of the “failure modes”; Furthermore, when the diagnostics probability is below a threshold, it is considered a “Novelty.” The Learning diagnostics machine bypasses the need for expert knowledge. The learning machine has a strong clas- sification power that can exceed the knowledge of human experts when a database of identified data is available. The diagnostic approach of the present invention combines supervised and unsupervised learning algorithms. Those algorithms allow adaptation to new identified data and improve the diagnostics performance. Fuzzy ART (Adaptive Resonance Theory) is an unsuper- vised diagnosis method. In this method, Supervised learning Neural Networks methodology such as SVM (Support Vec- tor Machine) MLFF (Multi Layer Feed Forward) present extremely powerful classification capabilities. This method eventually becomes the dominant method once enough identified data is accumulated to allow for a higher degree of reliability. By employing a combination of multi-method multi- feature automatic diagnostics the invention improves the confidence level of the generated alerts, and decreases the number of false alarms. Furthermore, the employed algo- rithm can be modified to reflect the maturity of the diag- nostic method. For example: higher weight is assigned to the expert system during initial development stages, while reduced weight is assigned to the same expert system once enough data is accumulated by the neural-network-based systems. A main diagnostics procedure combines all the classifi- cation results from all the diagnostics methods: over limits, aircraft sensors classifiers, snapshots classifiers, short term trend classifiers, long term trend classifiers and varying term classifiers, to a final diagnostics decision. The decision preferably includes more then one fault, as long-as these faults can potentially coexist. The main diagnostics proce- dure evaluates the decision confidence level using the clas- sification results and decisions history. A system in accordance with the present invention pro- vides a prognostic function in addition to the diagnostic function discussed above. In one embodiment, the system provides trend prediction by curve fitting. The curve fitting preferably employs a polynomial method. In another embodiment, the system employs an Auto Regression Mov- ing Average (ARMA) method to predict system behavior. In an alternate embodiment, the system also employs neural network prognostics. Two other prediction methods are fault prediction techniques. One is based on diagnostics of the predicted results, and the other is based on the faults history of a specific engine (engines’ sensitivity detection). The Polynomial method fits a polynomial for each existing parameter trend. The polynomial is then used to predict the future value of each parameter which leads to a prediction. The ARMA prediction model is applied by assuming that the vectors X1, X2, . . . , XR are samples of a stationary Gaussian process (see Box and Jenkins, 1974), and where Yt is t-th cycle (a moment of the time series) prediction, X. is i-th cycle vector, at is t-th day prediction error, K is a number of “history” moments, p is vector size, q is number of Moving Average (MA) parameters. The challenge is to find the unknown coeflicients ai, bl. from a “learning-set” of data. 10 15 20 25 30 35 40 45 50 55 60 65 20 A multilayer perception (one type of neural network) calculates an output as follows h K p Y1 = ¢ompuz[ E Vj¢hidden[2 Wp(lil)+iXitil + W0]], [:1 1 j:1 ‘: where q)(x) is a non-linear activation function, e.g. wl. MLP hidden layer coeflicients, v]. is MLP output layer coeflicients, p is vector size (dimensionally), h is number of hidden neurons, K is number of “history” moments, and the prediction task is to find coefficients vj, wl. Discussion of such networks and analysis can be found in Reference: Box, G. E. P., Jenkins, G. M., Time series analysis. Forecasting and control. (Mir, Moscow ’974) pp. 17406, which is incorporated by reference herein. Referring to FIG. 9, a block diagram is used to illustrate the decision process as applied to the extracted features. It should be noted that before commencing the decision process, sensors should be checked and data quality and validity considered. Because the decision process involves novelty detection and retraining with new detected features, it is important that the decision process is only performed on data that accurately represents engine conditions. As dis- cussed above, the sensor checking generally includes a validity check, a data quality check, a cross-check between correlated parameters, and a comparison between different engines on the same aircraft. For example, EGT and fuel flow are correlated and should deviate in the same direction. If the two parameters do not deviate in the same direction, a measurement or sensor problem may be the manifested fault rather than an engine problem. In one embodiment, these steps are undertaken during each flight phase over the course of a flight. The first step of the decision process is novelty detection (511, 521, 531, 541, 551), which is preferably performed separately for each group of features. For this purpose, features that characterize different phenomena are separately grouped. The groups include gas path snapshot features 510, vibration snapshot features (for each failure mode) 540, gas path trend (long and short term) features 520, 530, and vibrations trend features (for each failure mode) 550. Gas path features are grouped according to the failure modes that the features represent, such as snapshot 510 (abrupt failures) or trend data (developed failures), including long term and short term 520, 530. Vibration features are preferably grouped according to failure modes by employing the aggregated vibration fea- tures. Thus, each failure mode is associated with a group of vibration features that indicates its condition. The features in each group are features not used in the diagnosis of, nor indicated on, other failure modes. Consequently, each group is separately diagnosed as a snapshot 540 or as a trend 550. Diagnostic indices are preferably calculated at the pre- defined pointers in all relevant domains, as discussed above. Next, the system processes aggregated vibration features. Each feature is provided by aggregating all relevant har- monics in the corresponding domains. The vibration feature space relevant to each failure type thus contracts as a result of aggregating the vibration features. The novelty detection algorithms, shown by blocks 511, 521, 531, 541, and 551, detect an unknown combination of US 7,027,953 B2 21 group features in accordance with a classification strategy. A preferred classification strategy of system deterioration states for a diagnostic system, which improves and adapts its discrimination capability with a growing quantity of iden- tified examples for each state, is based on three stages. The first stage discriminates between normal operating and dete- riorated systems. It is presumed that in the first stage most of the examples are of normal operating systems. The classifier is thus able to detect normal operating systems and conversely initiate alerts when a new pattern is detected. The second stage discriminates between normal operating systems, a limited number of deteriorated stages, and new patterns (“novelties”). This classifier is preferably based on fuzzy ART algorithms, and is adaptable whereby the number of known patterns, or identified deterioration states, is increased as more examples are acquired. The third stage is preferably a robust classifier (based on neural network methodology), which can discriminate between a plurality of deteriorated system states. In order to increase system efliciency and reduce time-lag during diagnostics, the present invention employs a set of modular neural networks, which are combined in a voting system. This approach enables the application of neural networks for diagnostics or classification when the number of examples for each class is limited while the feature space is quite large. The number of examples needed for training a neural network is proportional to the number of nodes of the network, including the input layer or feature space dimension, and the inner layers nodes and the output layer which correspond to a number of classes. By enlarging the number of networks and dedicating each network to dis- criminating between only two classes (defect X or not) the resulting network architecture becomes simpler and contains less nodes. The examples needed for training a modular neural network can be reduced by factors of 107100 (depending on the number of classes and the feature space dimension) while maintaining similar detection rates and probabilities of false alarms. Several methods may be used for novelty detection. The nearest neighbors method, described in Bishop, C. M, “Neu- ral Networks for pattern Recognition” (Oxford University Press, NY. 1999), examines the Euclidean distance in the feature space between the current flight feature and all known features from historically representative flights. The feature space is the space in which the features for a specific classification scheme are provided. An example feature space is long-term shifts of ANl, AN2, AEGT, AFF. Fuzzy Adaptive Resonance TheoryiFuzzy ART, groups an ‘N’ dimensional feature space (flight history) into ‘N’ dimen- sional boxes. Each “box” represents a known operational area. If a combination of features from a given flight does not fall within any of the known boxes, it is considered a novel combination. Hybrid Systems 560, 542, 552, are used to combine the classification results from each group of features. The cor- relation between different types of features increases the reliability of the final decision. For example, decisions based on the gas path parameters are based on both snapshot features and trend features. A combination of different classifiers is used by combin- ing novelty detection with the classification to improve the False-Alarm-Detection-Rate ratio. For example, if the nov- elty detection algorithm indicates that the engine is not properly operating while the classifiers indicate specific faults, the confidence level is increased. On the other hand, if the novelty detection algorithms indicates proper engine operation while the classifiers indicate specific faults, the confidence level is decreased. 10 15 20 25 30 35 40 45 50 55 60 65 22 As shown in FIG. 9, different artificial intelligence tech- niques are used to classify the features of each group according to the related knowledge base. For gas path trends, where extensive a priori knowledge is available for a rule base, an expert system based on fuzzy or deterministic logic 522, 532, is employed. Furthermore, where informa- tion on probability distribution is available, Bayesian Net- works may be used. Expert systems employ classification methods that formulate the expert knowledge into a classi- fication decision, e.g. decision trees and fuzzy logic. Baye- sian Networks are statistical diagnostic tools. These two methods are well known in the art and are discussed in Cowell, R. G, Dawid, A. P, Lauritzen, S. L, and Spigelhalter, D. J, “Probabilistic Networks and Expert Systems” (Springer 1999), Chap. 10, in Bishopm, C. M, “Neural Networks for pattern Recognition” (Oxford University Press, NY. 1999), Chap. 2, and in Duda, O. R, Hart, E. P, and Strok, G. D, “Pattern Classification” (Wiley- Interscience Publication 2000), incorporated by reference as if fully set forth herein. As shown by blocks 512, 522, and 532, when enough identified examples are available, self-leaming algorithms are used. Examples of self learning algorithms include neural networks, including Multi Layer Feed Forward, Linear, Fuzzy ART, Fuzzy ART Map, and Radial basis neural networks. Other self-learning algorithms include Support Vectors, Decision Trees, Nearest Neighbor, and Fisher discrimination with or without Principal Component Analysis (PCA). Such algorithms are discussed in Bishop, C. M, “Neural Networks for pattern Recognition” (Oxford University Press, NY. 1999), pp 310. and Haykin, S. “Neu- ral Networks” (Prentice Hall, 1999), Chap. 8. For feature groups of vibration and vibration trends, hybrid systems 542, 552, are the preferred classifiers. Hybrid systems 542 and 552, provide an advantage by combining several diagnostic methods. Each group of vibra- tion features may have different properties, which are ame- nable to different classification methods. Nevertheless, the overall classification is combined from several groups of vibration features. The multi-classifier combination is thus a Hybrid system 560. Since each group relates to a specific failure mode, the principal component of the system is an expert system based on Fuzzy Logic that reflects the rea- soning of the vibration analysis expert. The component analyzing the severity of each failure is implemented via self-learning systems (i.e., MLFF NN, and Fuzzy art). These systems correlate between vibration features and failure severity. Referring back to block 470 of FIG. 5, all decisions of each classifier are trended. The trends are then analyzed and a special classifier subsequently processes their features. The results of the classifiers for related groups of features are combined in block 560 of FIG. 9. A decision is then made in block 570, once on current data, and a second time on predicted data, to arrive at a current diagnosis as well as a future prognosis of the mechanical system. This decision 570 preferably includes a confidence level determined in block 590. The confidence level of the classification 590 is preferably arrived at by combining information related to data quality, feature quality, cross validation between classifiers, cross validation with previous diagnostics results, and each classifier confidence level. In one embodiment, each classifier is associated with a correspond- ing mechanism for generating the classification confidence level. For example, in fuzzy logic, the confidence level depends on the membership function combined with the specific rule confidence level. The decision arrived at in US 7,027,953 B2 23 block 570 is also considered as part of a trend of decisions in block 580. The previous 375 system decisions are exam- ined to determine if the arrived at decision of block 570 is indicative of a fault or condition that becomes apparent when observing a series of decision as a continuum. In addition to the three types of Trend features (Long, Short and Varying terms), the Expert Systems also use peripheral information on the cycles, indicating engine load, maintenance, and comparison to fleet statistics (from similar engines). Some of the peripheral parameters are relevant to the engine condition evaluation. For example, if an engine is new, or just after total overhaul, the probability for fatigue problems is lower than for an engine that has been in use for 20,000 cycles. Another example is the comparison to fleet statistics. If the statistics of the fleet indicates that after 25,000 cycles, a specific problem appears, then the prob- ability of the diagnostics for this problem increases. The prognostics of specific over limits are preferably based on multi-curve fitting, extrapolation, and a prediction based on parametric models. Multi-curve fitting is provided by fitting a multi linear curve to each trend parameter such that the error corresponding to each curve is less than a threshold. Next, extrapolation is performed by using the best fit curve to predict future feature data. Finally, the prediction step uses the data provided by the extrapolation to predict the trend behavior and generate an alert, if so indicated. The system further employs parametric models such as ARMA and polynomial fitting, to allow on-line detection and initiation of immediate alerts. For example, a stall in the compressor manifests in pressure pulses in one or more stages of the compressor. The pressure pulses rate depends on the rotating velocity of the compressor shaft but the frequencies are not proportional. This phenomenon can be detected using measurement of the compressors case vibra- tions that can sense the pressure pulses. During stall, the vibration signature exposes the pressure pulses frequency and the higher harmonics. Higher vibration levels over all the frequency range (with the regular pattern) are observed by modulation of the pressure pulses frequency with the blade pass of the compressor stalled stage. The feedback either from experts or from the field (maintenance data) is used for reinforcing the decision schemes. There are two general cases where feedback is employed. First, feedback upon alerts on “known states” either reassuring or contradicting the specific automatic diagnosis is used to recheck or fine-tune the feature extrac- tion or the decision process algorithms. Second, feedback upon a diagnosed “novelty” is used to enlarge the knowledge base. The specific failure mode is analyzed and the appro- priate features and patterns are used for retraining the decision process algorithms. After completion of the retraining, the specific anomaly becomes a “known” defect that can be automatically detected. The retraining process depends on the specific decision or classification algorithm. For supervised learning techues (neural networks) the training process is initiated when adding the new patterns to the training set. For expert systems new rules are added to the rule base. In the case of Fuzzy ART algorithms the appropriate class is identified. After the retraining of the system, sets of new parameters for the decision process algorithms are obtained. After the update of the airborne system or the ground station configuration, the system is able to automatically identify the new defect. Example System Operation The operation of a system in accordance with the inven- tion will now be illustrated by reference to the operation of 10 15 20 25 30 35 40 45 50 55 60 65 24 the system when diagnosing the condition of a main bearing in an aircraft engine. The description will refer to data provided in Tables A, B, and C, which illustrate failure mode data for select components of an aircraft engine. The values in the “Pointer” column, provided in Tables A and B, are coeflicients used to arrive at an observation frequency from a shaft frequency. The reference shaft frequency is provided by a code in the “RPM code” column. The RPM codes refer to codes from Table C, designating various frequencies in the observed engine. For example, code 2 refers to the rotation frequency of the High Pressure shaft. The pointer value from Tables A and B is multiplied by the frequency from Table C to arrive at the observation frequency. With reference to Table A, the “Harmonics” column designates the number of Harmonics observed for the cor- responding fault. The “Channel” column provides a data channel designation, associated with one of several vibration sensors of the system. Accordingly, Table A identifies the sensor data which will be used to generate signatures for each fault. The weight corresponding to the channel desig- nates the decision significance of features arrived at from the channel data. For example, a higher significance assigned to Channel 1 over Channel 2 will entail that features from data provided by Channel 1 will contribute more significantly to the final decision than features from data provided by Channel 2. The vibration signatures of a system contain information related to the structural properties, dynamic properties (rotation related) and other properties assumed to be aero- dynamic related. The structural properties of different com- ponents are manifested through the natural (modal) frequen- cies. The dynamic properties are manifested as energetic peaks at frequencies correlated with their rotation speed. The aerodynamic properties appear at certain operating modes of the system and their manifestation is usually dependent on the specific system. The first required step for a comprehensive vibration analysis is the identification of each manifested character- istic in the signature. Each class of properties (structural, dynamic and aerodynamic) that is reflected in the vibration signature will require different type of information for its identification. In order to identify the structural properties manifested in the signature, detailed information on the natural frequencies (of all the relevant modes) and their dependency on the rotating speed (or as a substitute the geometry, material, and physical properties) of the engine components is required. Dynamic properties identification requires only geometric and engine general structure infor- mation. The identification of aerodynamic properties would require a comprehensive investigation of the operating modes of the subject system. According to decision process stages illustrated above with respect to FIG. 5, the first, preliminary step, is the failure mode analysis 480, which includes understanding the failure mechanism and its effects on the vibration signature. The automatic algorithms and/ or the human expert should be aware of the qualitative as well as quantitative type and range of change in the signature pattern that should be expected. For structural changes (e.g. blade tip degradation, cracks curling etc) the importance of recognizing and understand- ing the exact failure mode mechanism and its manifestation in the vibration signature is reinforced because it can be easily confused with other structural changes. Other struc- tural changes may cause the excitation of a “new” natural frequency to appear, or changes that will cause a shift in the natural frequency. All types of structural changes generate a US 7,027,953 B2 25 change in the natural frequency pattern. The estimation of the type of pattern change, its dimensions and its location (which of the modes are affected) requires a preliminary investigation for each part and each structural change type. For a failure related to the dynamic class, the change in the signature is directly related to the rotating motion (e.g. bearings, gears etc), the structural and geometric informa- tion on the engine and specific components is sufficient for estimation (prediction) of the expected changes in the sig- nature. These types of failures generate distinctive peaks at precise locations (pointers) in the different analysis domains. For the purposes of this illustration, an example bearing is said to potentially have four failures: wear of the inner race, wear of the outer race, faulty roller, and wear of the cage surface. This illustration will discuss the relevant observations for detecting a bearing defect of the outer race, such as by wear of the outer race. When the worn out surface of the outer race is in contact with other surfaces of the bearing, the increased friction between surfaces produces mechanical shocks. The rate of shock production depends on the number of rollers and the diameters of the inner and outer races of the bearing. This information is used to determine how the failure will mani- fest itself as observed by the system. Because the failure will manifest itself at any rotating speed, the system of the present invention does not require specifying a system state or a minimum load for triggering a diagnosis. Second, the sensor location is selected as a point with high transmission of vibrations to the bearing. In this example, this position is the engine case opposite to the turbine. The vibrational observation is expected to be a series of shock indications that are proportional to the rotating speed of the bearing and corresponding shaft. A spike is expected to appear in the frequency domain spectrum, at a frequency corresponding to the shock rate. The spikes in the frequency domain should include a first peak at a first frequency and a series of additional peaks at integer multiples of the first frequency, which are the associated harmonic frequencies. The conclusions of the failure mode analysis are applied for selection of the preferable measurement setup and data analysis procedure. As for other types of failures, signatures of a normal operating system (in good condition) as well as of defective systems (with the specific investigated fault) should be collected. Those signatures (resulting from the signal processing of the measured raw data) provide the database that would be used for tuning the fault identifica- tion algorithms (classification). The next step includes calculating or observing where the spikes in the frequency spectrum would appear. The location of the spikes on the frequency spectrum provides the rel- evant pointers for the fault. Table A provides relevant pointer data for the bearing. As indicated in Table A, the frequency where the spikes appear depends on the geometry of the bearing and the rotation speed of the corresponding shaft (indicated as lb501). Table C provides the reference rota- tional value to employ in computing the pointer locations. The harmonics of the first frequency, particularly the odd integer multiples, are also taken into consideration as point- ers (indicated as lb502 . . . lb4o7) Since the nonlinearity of the phenomenon provide mainly odd numbered excited harmonies. If the outer race of the specific bearing rotates, the side bands of this pointer are also observed. The side- bands are usually generated due to modulation of the shock amplitude, amplitude modulation, with the rotating speed of the outer race. The side band pointer frequency is calculated by adding and subtracting the sideband rate, or the outer race rotating speed (N2) to and from the main pointer frequency 10 15 20 25 30 35 40 45 50 55 60 65 26 (lb501). Another parameter is the number of sidebands expected to be observed around each pointer. From experience, five to seven sidebands are considered for bear- ings. The im sideband frequency is calculated by adding and subtracting the product of i and the sideband rate to and from the main pointer frequency (lb501). The domains where the pointer data should be observed are selected for the component. Table B illustrates the selected domains and indices for select components of an aircraft engine, including the main bearing. For bearing data, the rate of the shocks is not constant due to variations in shaft rotating speed. To overcome this variability, the data is correlated to the shaft rotating speed during collection to provide a mapping from the time domain into the cycles domain. This provides a constant shock rate and after a second transformation into the orders domain (the spectrum of the signal in the cycles domain), a series of spikes at the corresponding order is obtained, i.e., a number of contacts or shocks per revolution, including corresponding harmonies. When the engine bearing is in good condition, peaks are not expected at this specific order. Because a series of spikes is expected in the orders domain, in the quefrency domain, an increase of the Cepstrum amplitude is expected at the respective quefrency, which is the reverse of the spikes rate in the orders domain. In the orders of the envelope, the envelope signal is always calculated in the cycles domain, where the same phenomenon is expected. The Cepstrum of the envelope orders will reflect the same behavior as the Cepstrum of the orders. In the phase average, which is also calculated in the cycles domain, the shocks will be canceled because their rate is not an integer multiplier of the rotating speed and therefore their phase varies in each cycle. The feature extraction step, based on the signatures database, includes comparison of the actual signatures to the appropriate baseline signatures of the system in the same operational conditions. The comparison between the pat- terns of the actual and “normal” signatures is achieved by calculating a set of diagnostic indexes for each pointer. The diagnostic indexes (Vibration Features) characterize the change in the signature pattern around a predefined region (Pointer) that is associated with a specific failure mode. The comparison is performed by reference to Baseline signature obtained from the “normal” signatures available in the database. The Baseline signatures and the determination of their tolerances is one of the key factors for a successful diagnostics. All vibration features related to a specific failure mode (at all relevant harmonics from all analysis domains of all diagnostic indexes), are aggregated to a figure that repre- sents the probability of occurrence of the specific deterio- ration severity. The aggregation stage requires a significant amount of signatures, both of normal operating and of defective systems, for determining the appropriate scaling of the diagnostic indexes for each fault type. As for all other engine features, the vibration features are trended and more features are extracted from the trend analysis process. Next, an analysis of the observed data is designed so as to detect the fault when it arises. Note that all possible data analysis, in all domains, is performed for the observed data since the analysis is not specific for the detection of a single fault type but is often employed for diagnosis of more than one component in the system. As shown in Table B, the domains of interest for the engine bearing outer race failure are: Orders, Cepstrum, Orders of the envelope, and Cep- strum of the envelope. These domains are preferably calcu- lated after synchronization with the rotating speed of the corresponding shaft, as discussed above. US 7,027,953 B2 27 The feature extraction process consists of comparing the actual signature of the system to the baseline signature of the same type, reflecting the same conditions or flight regimes. Both the baseline and actual signatures are calculated using the same analysis procedures. The comparison is performed by calculating a set of diagnostic indexes at each predefined pointer and its sidebands as well as for all harmonies of the pointer and their sidebands. The diagnostic indexes indi- cated for each pointer in the failure pattern are calculated for all relevant harmonies of the pointer and its relevant sidebands, in each relevant domain, as indicated in the failure mode pattern provided in Table B. As indicated in Table B, for the ingine bearing outer race defecr detection, the diagnostic indexes of Amn, Gmn, Hmn, Mfrms, So, and Rdo are calculated in the Orders and 10 28 Envelope orders domainds at the first 3 odd harmonies and at the first 3 sidebands. Finally, all diagnostic indexes belonging to a particular failure node are aggregated using their relative weights, as provided in Table A, in the relevant domains. This aggregate feature provides an indication of component health for the analyzed fault condition. Hence, for the main bearing outer race failure, the aggregate feature would indicate the likelihood that the bearing outer race is failing. In one embodiment, this indication is by observing that the aggregate index result exceeds a threshold level. The foregoing merely illustrates the principles of the present invention. Those skilled inthe art will be able to devise various modifications, which although not explicitly desribed or shown herein, embody the principles of the invention and are thus within its spirit and scope. US 7,027,953 B2 29 30 H I u u u —.v m- :93. "9‘ or“). a. , . Jun..." | 'Wr'fir': if" in .'.ii. 1L.-..':nm in; '... 1. Appendix A 1.1. Zero Crossing [Stat,zct,fN1,fN] = zeroCVec(V ec,SampRate,maxChange,NPointsSmooth,DCMeth,MultF actor) This function calculates the frequency of a wave as a function of time using the time instants of zero crossings of the wave. Received parameters: 0 Vec — time history vector (raw data vector or another vector) beginning at Offset of length Length points: vector of integers of length 256-500000 samples. SampRate - Sampling rate. maxChange is the maximum fi'equency change . NPointsSmooth - number of points for averaging to the smooth function. (Default=10*lOl). DCMeth - flag which indicates whether to subtract DC (=1) or not (=0). (Default=1). - MultFactor -fN/le multiplication factor. (default=1). Returned values: Stat - status / error code -1 = in case of wrong number of arguments. -2 = in case of wrong arguments. zct the time vector of starting points of periods for the Xtth/fNI . fN 1 the smooth of fN. - IN the calculated frequency vector. Algorithm: 1. Check input data. 2. Calculate time vector of the zero crossing of Vec. (zct). 3. Calculate the instantaneous rotating frequency, fN. = +0 = l,2...length(zcz)) 2C - I 4. smooth frequency vector fN, jN1=smooth(flV). 1.2. Smoothing [y_smoothed, Status] : smooth(Y, window_length). ‘ V This function calculates the moving average of Y. The averaging is applied to +/.- 0.5 moving window length. Received parameters: Y — input data vector window_length - moving average window. Returned values: y_smoothed - Y alter moving average Status - debugging status: 0 - the function is working correctly 1 - Warning: all the values in Y are zeros Us 7,027,953 132 31 32 <0 - error code. Algorithm: 1 Check input data. Forcing window__length to be odd. 2. 3. N = mindwindow_ length} Iength(Y)) . 4. Calculating the moving average by convolution of Y with vector of the same length, M . which contains the l/ N ’s. conv(f,g) : If(x—y)*g(y)dy , In our case it would be: min(k,N) y _smoothed(k) = Z 1 /N*Y(k +1—j) j=max(l,k+1—N) k = l,2,3,...,2 * N -l 5. Special care for the edge points, for the first points: 1 2‘1: -l y_smoothed(k) = 2W4 Y(i) k =1,2,...,H 2 1.3. Background Spectrogram [Statfiackground]=SpectrogramBackground(Vec,zctNl,N1,harrnonyNl,deltal ,zctN2,N2,harmo nyN2,delta2,NewSRate); . and in a similar way for the last J points. Received parameters: Vec - time history vector (raw data vector or another vector) beginning at Ofl‘set of length Length points: vector of integers of length 256-500000 samples. zctNl — time vector for the N1 (same length of N1). N1 — Rotating fi'equency vector in rotations per second [Hz]. harmonle — vector of the number of harmonies of N1 we want to “clean”. (floats). deltal — the width of the frequency stripe (N!) which we want to filter (+- delta] for each direction). zctN2 — time vector for the N2 (same length of N2). N2 - Rotating frequency vector in rotations per second [Hz]. harmonyNZ - vector of the number ofhannonies of N2 we want to “clean”. deltaZ — the width of the fi'equency stripe (N2) which we want to filter (+- deltal for each direction). ' I NewSRate ~ Sampling Rate (could be the new Sampling rate if zooming was called). Returned values: Stat — status / error code (integer). - Background — time history vector, afier removing (by dynamic filter) the rpm and its harmonies. US 7,027,953 B2 33 34 Algorithm: 1. Check input data 2. Filtering all the given harmonics of N1 by dynamic BP._ 3. Filtering all the given harmonies of N2 by-dynamic BP. sin[Q‘ ('t—t )] _ sin[Q(t-ti )] _ P '- fume)—Zf(t.)——§T—Zf(ti)———Qt(t_,i) ‘ Qt—t. ‘ ( 1) Q Qe = envelope fiequency p = Q/ 9g 1.4. Re-sampling [Stat,Nvec,Tvec,newNvec,newTvec] = re-samplingVec(originVec,Offset,Length,SampRate,CpS,XCpS,fac,FiltParStruct) This function re-samples the signal using a changing rate proportional to the rotating frequency as measured by N1 or N2. Afier re—sampling of data: 0 The re-sampled vector is represented in a new domain '— cycles domain (pseudo-time domain) and after Fourier transform it will be mapped in the Orders domain (pseudo- frequency domain). 0 Thenew sampling rate in the cycles domain is equal to the multiplication factor of the rotating frequency (Fac), such that one period of revolution — one cycle (which becomes constant afier re-sampling) contains Fae sampling points that corresponds to 1 cycle. 0 The maximum Order (revolution rate multiplier in the pseudo-frequency domain) in the re—sampled signal depends only on the new sample rate or no. of sampled points per revolution, i.e. F 210/2. 0 The Order resolution (pseudo-frequency domain resolution) depends on the no. of cycles considered for FFT calculation, i.e. AOrder-‘l/NoCycles. Received parameters: - FileName - time history file. 0 Offset — start point in the raw data matrix (long integer). Length - length of vector Vec (long integer). CpS — Rotating frequency vector in rotations per second [Hz] (vector of length Length/SarnpRate*20). - XCpS - time vector for the CpS (same length of CpS). o Fac - multiplication factor for calculation of the instantaneous sampling rate (float). o FiltParStruct - structure for filtering options containing: 0 FiltType — Code for filter type: LP, HP, HP, BS (integer). Freql - first cutoff frequency (float). Freq2 - second cutoff frequency (float) - only for BP and BS filters Slope - filters ‘slope in [dB/Oct] (integer). PdB - Level at cutoff frequencies [dB] (integer). 0000 3 US 7,027,953 B2 35 36 .43 ‘n-‘m '2. n n in ""5!- “‘5r vii emf-17w r1 r1» 7”"): . i l‘ .J' In .JL. 1132:. .33“ Lil mew-It; o PhaseType - filter phase distortion code: with/out (integer). Returned values: Stat - status / error code 0 Nvec - vector of synchronized time history (float vector of length dependent on the new changing sample rate). - Tvec v new time scale, after re-sampling. Algorithm 1. Read / build the data vector Vec beginning at Offset of length Length. 2. Linear interpolation /extrapolation of CpS(XCpS) for time values of t=[0:1/SampRate:Length/SampRate]. The resulting vector is: Freq = interp1(XCpS, CpS, t, ‘1inear’). Far: ‘ 3. Digital integration of Freq vector: Phi, (pl. = ————-—— Freq,‘ . SampRate h, The vector Phi is now a function of t. 4. Linear interpolation of t(Phi) for values of Phinew=[0: l :ftx(max(Phi))]. The new time values founded by interpolation are t1 ( t1 = interpl(Phi, t, Phinew, ‘1inear’) ). 5. Filtration of Vect(t) as required by FiltParStruct - always without phase distortion. 6. Cubic interpolation of Vec(t) for values of t]: Nvec = interpl(t, Vec, tl, ’cubic’). 1.5. Phase Mean [Stat, Pthec, xVec] = PhaseMean(Vec, OffSet, Length, Noofllevs, SampPRev, NoofFrames, FiltParStruct) ' Average of a number of periods of rotation (time history — raw data) previously synchronized to l/rev signal (Resampling) of N1 or N2. The time history result of phase averaging contains only the harmonic components of the original signal that are proportional (by an integer multiplication factor) to the rotating frequency. All other elements in the original signal are cancelled by this operation (depending on the number of periods averaged). Received parameters: 0 Vec - time history vector [raw data vector afier synchronization (e.g.Re—sampling, or another synchronized vector)] beginning at Offset of length Length points: vector of integers of length 256-500000 samples. Offset - start point in the raw data matrix (long integer). Length - length of vector Vec (long integer). NootRevs - number of revolutions in each set (integer). SampPRev - number of samples per revolution (integer). NoofFrames — number of frames to average (integer). FiltParStruct - structure for filtering options containing: FiltType - Code for filter type: O-no filter, l-LP, 2-BP, 3—HP, 4-BS (integer). US 7,027,953 B2 37 38 if}- "WE-#3??? .1. "REIRUUE' Freql - first cutoff frequency (float). Freq2 - second cutoff frequency (float) - only for BP and BS filters. Slope - filters slope in [dB/Oct] (integer). PdB - Level at cutoff frequencies [dB] (integer). PhaseType - filter phase distortion code: with/out (O-flltfilt,l~filter). Returned values: - Stat - status / error code (integer). - Pthec - phase averaged time history vector(float vector of length=Nooflievs*FrameSize). ' , o xVec - pseudo-time vector (float of length=NoofRevs*FrameSize). Algorithm: . It is assumed that the given Offset points to the beginning of a revolution of the shalt. The steps of the algorithm are: 1. Check input parameters integrity: a. Length>=SampPRev*NoofRevs*Noo£Frames b. Offset+Length < length of vector Vec c. FiltParStruct.PhaseType=l (without phase distortion) 2. Calculate the average of the time history in frames of SampPRev*NootRevs for N y(t,l ,...t,-M ,...t,-m) Nooflfiames times: Eph(t,,...zM,...tKM) = ~———————~——’ ' N , where: M=SampPRev, K=NoofRevs, N=Noofl7rames, y(t)=Vec. 3. If requested, apply the appropriate filter on the averaged data. For re-sampled data the sampling frequency is 1 (corresponding in the time domain to SampPRev sec per revolution). The filter frequencies have to be entered as orders/Nooflirames. ll 1. 6. PSD [Stat, Cvec, xVec, NOverLap] = myPSD (Vector, Offset, Length, FrameSize, FFTSize, SagnpRate, Window, Overlap, Units) Calculate PSD of the signal and enable overlapping. Received parameters: 0 Vec - time history vector (raw data vector or another vector) beginning at Offset of length. Length points: vector of integers of length 256-500000 samples. Offset - start point in the raw data matrix (long integer) Length - length of vector Vec (long integer) FrameSize - frame size for calculation (integer) NootFrames - no (number) of frames to average (integer) FFTSize - frame size of the FF T output vector*2 - enables zero padding (integer) SampRate - Sampling rate of the raw data (float) Window - code for window type to apply on vec (integer) 0 = "Rectangular" 1 = "Hanning" (default) 2 = "Hamming" 3 = "Blackman" o Overlap - Overlap percentage (float in the range 0-100 (0 default)) 5 US 7,027,953 B2 39 40 0 Units - code for physical units (type) of the output spectrum (integer) 0 = (G‘Z/Hz) (default) 1 = (G) 2 (dB) 3 (velocity m2/sec2/I-lz) 4 (velocity [m/sec]) 5 (velocity [in/secD Returned values: 0 Stat - status / error code (integer) o Cvec - spectrum vector (float vector of length = FFTSize/Z) o xVec — vector of frequencies/orders corresponding to the Cvec (float vector of length = FFTSize/Z) o OverLap - number of frames. Algorithm: N Z|FFT(yi A w)!2 PSD — Power Spectral Density is defined as PSD(y) = where: y.- — the time history of y in the ith frame, w 7 window function, At" — spectral resolution, N — number of frames. 1. 7. Envelope function (Stat, Eancc] = envelop(FileName, Channel, Offset, Length, FrameSize, DCMeth, SampRate ,FiltParStruct) Calculates the instantaneous / mean envelope of the absolute value of a time history (raw data). Received parameters: FileName - time history file. Channel - the channels we are interested in. Offset - start point in the raw data matrix (long integer). - Length - length of vector Vec (long integer). o FrameSize - frame size for calculation (integer). 0 DCMeth — code of method for DC removal/calculation(integer). Options: l-per frame. 2-Calculated over all Vec. 3-no DC. SampRate — Sampling rate of the raw data (float). FiltParStruct - structure for filtering options containing: - FiltType - Code for filter type: O-no filter, l-LP, 2-BP, 3—HP; 4-BS (integer). Freq] - first cutoff frequency (float). Freq2 - second cutoff frequency (float) - only for BP and BS filters. Slope - filters slope in [dB/Oct] (integer.( PdB — Level at cutoff fi'equencies [dB] (integer). PhaseType - filter phase distortion code: with/out (integer.( Returned values: Stat - status / error code. US 7,027,953 132 41 42 l’ a n; w, a”, a.» M’W "w- vi"'h lr" e Jr r 1.4 “31.11:. gu- Lil mi: Eanec - envelope history vector (float vector of length equal to Length/FrameSize). Algorithm: The envelope of a time history (previously BP filtered or not) is defined as: Env( y) = lH {F i123). ( y)}| where: y — time domain vector, H {0}— Hilbert transform defined by: r20) = Home; ZY (f ), f > 0 Ya (f) = Y( f), f = 0, Y( f) = F{y(t)} the Hilbert transform ofy will be: 0, f < 0 ya) = Imlr' [1: ml}. mt) = y(t)+fi(t) For calculations of the envelope, the absolute values of the complex vector resulting from the Hilbert transform should be calculated. When the input parameter FrameSize<Length, alter the envelope calculation the frames of the resulting vector should be averaged term by term and the resulting average envelope vector will be of length FrameSize. When F rameSize=Length, the output vector represents the instantaneous envelope and is of a length of FrameSize. 3}“) dr and it can be proved that for: — r n 1.8. Cepstrum [Stat, Ceps, quefrency] = Cepstrurn (PSDvec, FF TSize, SampRate) Calculates the absolute Cepstrum of a time history (raw data or another vector in the time domain). Cepstral analysis is a nonlinear signal processing technique that is applied most commonly in speech processing and homomorphic filtering [1]. Received parameters: PSDvec - PSD vector. FFTSize - frame size for calculation (integer, power of two. SampRate - Sampling rate of the raw data (float). Return values: 0 Stat - status / error code (integer). 0 Ceps - cepstrum vector (float real/complex vector of length = FFTSize/Z) quefrency — vector of quefrencies corresponding to the Ceps (float vector of length = FFTSize/Z) The real cepstrum is the inverse Fourier transform of the real logarithm of the magnitude of the Fourier transform of a sequence: RC)! = Small?" {JnGFbIUflDll The real cepstrum is a real-valued-function. ‘ [y,ym] = rcepstx) returns both the real cepstrum y and a minimum phase reconstructed version ym of the input sequence. y = real(iffi(log(abs(ffi(x))))); 1.9. US 7,027,953 B2 43 Diagnostic Indexes - Faultind [Stat, Amn, Gmn, Hmn, Mfims, Rdo, So, Rk, Cm, Re] = faultind(P, PReff, Dfl, DfReff, FStart, FEnd, PUnits, flag) This function calculates 9 diagnostic indexes of a spectrum compared to a reference spectrum. The indexes evaluate the differences between a certain spectrum and a reference spectrum (the appropriate base line in our case). Received parameters: P — Vector of the spectrum to diagnose (floating vector) PReff - Vector ofthe reference spectrum (floating vector - base line) Dfl - frequency resolution of P (float) DfReff- frequency resolution of Preff (float) FStart/F — starting frequency for comparison (float)/main frequency for comparison (in case flag=1)(float) FEnd/Npoints - end frequency for comparison (float)/how many points from each side (in case of flag=1)(integer) PUnits - Code for Units of both spectra (integer l-FFT O-PSD) flag — flag=l means that the user sent F & Npoints. (Default = 0). Returned values: Stat - status code (integer) Amn - difference of Am of spectrum P and the reference spectrum PReff (float) (Anni-Arithmetic mean) Gmn — difference'of Gmn of spectrum P and the reference spectrum PReff (float) (Gm-Geometric mean) Hmn — difference of Hmn of spectrum P and the reference spectrum PReff (float) (Hmn-Harmonic mean) Mfrms - Mfrms index of spectrum P compared to the reference spectrum PReff (float) (Mfrms-Matched Filter Root Mean Square) Rdo - Rdo index of spectrum P compared to the reference spectrum PReff (float) (Rdo-rms of spectral difference) So - So index of spectrum P compared to the reference spectrum PReff (float) (So-Sum of Squares of Differences) Rk - Energy Ratio. Cm - center of mass. Re - relative entropy. US 7,027,953 B2 45 ilnllo‘ Amn(Y) =2010gl0 N , AAmn = Amn(Y) — Amnflflqr) i" [05 220mg", J5 m——] Gmn(Y) = AGmn = Gmn(Y) — Gmnmi) N*10"5 ("I l IY.“ ] lYnalz Hmn(Y) = 2010g10 , AHmn = HrmKY) — Hmn(Y,e[) 1 "~ Mfrms =1010g,0[7v—Z Rda= fi:(2010g10011l)—ZOIOgIoQYuAV 46 US 7,027,953 132 47 48 1.10. Statistical moments - Moment (Stat, Mvec, maxRMS] = momentT(Vec, Offset, Length, MomOrder, FrameSize, LenFrameSize, SampRate, DCMeth, FiltParStmct) This function calculates the different moments of the probability distribution of the data as a function of time or as a function of RPM: DC, RMS, Slcweness, and Kurtosis. Received parameters: - Vec - time history vector (raw data vector or another vector in the time domain) beginning at Offset of length Length points: vector of floats of length 256-500000 samples. I Offset - start point in the raw data matrix (long integer) - Length - length of vector Vec (long integer) - MomOrder — code for moment type to calculate (integer): l - DC 2 - RMS 3 — Skweness 4 — Kurtosis - FrameSize - flame size for calculation (integer) or a vector of different frame sizes(when the moment is calculated as a function of RPM) 0 LenFrameSize - length of FrameSize. - SarnpRate - the sampling rate of the raw data (float). o DCMeth — code of method for DC removal/calculation(integer). Options: l-per frame. 2-calculated over all Vec. 3—no DC. 0 FiltParStruct - structure for filtering options containing: FiltType 7 Code for filter type: O-no filter, l—LP, 2-BP, 3-HP, 4-BS Freql - first cutoff frequency (float) Freq2 - second cutoff frequency (float) ~ only for BP and BS filters Slope - filters slope in [dB/Oct] (integer) ’ PdB - Level at cutoff frequencies [dB] (integer) PhaseType ~ filter phase distortion code: with/out (integer) 0 ~ filtfilt l — filter Returned values: 0 Stat - status / error code 0 Mvec - probability distribution moment history vector (float vector of length equal to Length/FrameSize) - maxRMS — maximum(RMS). (float). Algorithm: The function will calculate the specific moment as a function of time ( FrameSize = Length / max(k) ) or a single value for all the samples of the vector (FrameSize = Length). 10 US 7,027,953 B2 49 50 . “an-3y .afl .1 _. ‘ m ' ..r ...r .. ...t is: 3:39;: The calculation steps are: 1. Check validity / integrity of inputs. 2. If FrameSize<Length (many times), calculate the no. of frames. For each frame: 2.1. Read the appropriate frame of Vec. 2.2. If requested apply the appropriate digital filter to the frame in Vec. The filter should be applied on a longer vector than the requested F rameSize for moment calculation,'i.e. 1024 sample points (if available) preceding the actual frame offset should be concatenated before filtering and ignored afier the filtration. 2.3. On the filtered data calculate the requested moment. The DC term in all the following equations refers to a value predetermined depending on the parameter 2y; DCMeth. The definitions of the moments are: DC,(y,.MM.M ) = where: y:Vec, N=FrameSize, k=l,.. .,N’, N’=Length / FrameSize inf—D64»): 1 :94 N EG,—DC.(y)l i=1»: Skew = . . = ___._.__ ness(y) 16k (yn.,,..1km) N ' EU,- —DC.(y))" j'ikol K t S. = . . _ .___.___—_ ur 0 150’) 7g (XMHM) N'RMsfly) 1.11. SegRPfl/i [Stat, NRPM, Frames, T, Starts] = segRPM( RPMvec, tRPMvec, DRPM, SampRate, minRPM, maxRPM, minimalFrame) This function calculates a vector of frames and the corresponding segments of time with approximately constant rotating speed, i.e. time segments in which the rotating speed is within predefined boundaries. Received parameters: 0 RPMvec - vector of RPM values in [Hz] 01' [RPM] or (float vector of length 256-50000 samples) 0 tRPMvec - vector of time instants corresponding to RPMvec values. If the RPMvec is sampled at constant rate tRPMvec is of length=2 and contains the start time and the sampling rate of tRPMvec. Float vector of length 2 or = length(RPMvec). o DRPM - the allowed range of the RPM in each frame (iDRPM) in same units as RPMvec. o SampRate - sampling rate (in order to calculate time frames). 0 minRPM — minimal value of RPM range. 11 US 7,027,953 B2 “3 9-7“ -:_ 9| )1 u "ml! “ . ‘5‘: --.-* ' . 1.. ".11. 1L." ..::.. .n‘fir‘fix' .v" n? m E12? 0 maxRPM — maximal value of RPM range. - minimalFrame - uses for warning(Stat=1). default=1000. Returned values: 0 Stat - error code 0 NRPM — Vector of average RPM in each frame of the Frames vector 0 Frames — Vector of frame sizes for the vibration time history. The RPM is within the boundaries defined be DRPM in each Rama. 0 T - vector of time start instants of each frame in Frames. I Starts - vector of the frames's starts indexes. Algorithm: The function searches for distinct segments of time in which the RPM is within a certain range. The stages are: 1. Check integrity / validity of the data. 2. Starting from the smallest RPM and up to the maximum value, build a vector of mean RPMs for each step: '0 = 0; for ii = 1 : (max(RPMvee) - min(RPMvec)) I2 IDRPM. - ind1 = find (RPMvec >= min(RPMvec) + (ii-1) ' 2 ' DRPM 8: RPMveC < min(RPMvec) + ii " 2 " DRPM); if isempty(ind1)~=0, indd=dift([1 indlp; i0=i0+1; inds=[1 find(indd>1)]; inde=[find(indd>1)-1 length(indd)]; NRPM(i0:iO+|ength(inds)-1) = min(RPMvec) + (ii-1) ' 2 ' DRPM + DRPM; Frames(i0:i0+length(inds)-1) = fnd1(inde) —ind1(inds) + 1; T = [T ind1(inds)/SampRate] ; End End 1. 12. Spectrogram [Stat, Smat, fVec, tVec] = Spectrogram(Vec, Offset, Length, SampRate, FrameSize, FFTSize, NootPSDs, Window, Overlap, Units) 12 US 7,027,953 B2 ,4, “mg, n in if u u "up w; i'. in, ‘9 I mucus-u ..'.._ ,..u ‘11““1rar‘ :m m -1113: .13: Calculates the Spectrogram of a time history (raw data or another vector) enabling zero padding before FF T calculations (improves the frequency resolution) and overlapping of frames for improved time resolution. Received parameters: 0 Vec - time history vector (raw data vector or another vector) beginning at Offset of length Length points: vector of integers of length 256-500000 samples. Offset - start point in the raw data vector (long integer) Length — length of vector Vec (long integer) SampRate - Sampling rate of the raw data (float) FrameSize — frame size for calculation (integer) FFTSize - fi'ame size of the fit output vector*2 - enables zero padding (integer) Noofl’SDs - no of spectra to calculate (integer) Win - code for window type to apply on vec (integer) 0 = "Rectangular" l = "Hamming" (default) 0 2 = "Hamming" = "Blackman" - Overlap — overlap percentage (float in the range 0-100 (0 default» 0 Units — code for physical units (type) of the output spectrum (integer) O = (GAZ/Hz) (default) 1 = (G) 2 (dB) 3 (velocity m2/sec2/Hz) ’ 4 (velocity [m/sec]) 5 (velocity [in/seep Returned values : 0 Stat - status / error code (integer) o Smat - spectra matrix (float matrix of size = [FFTSize/2 x NootPSDsD - fVec - vector of frequencies/orders corresponding to the Smat (float vector of length = F FTSize/Z) 0 tVec - vector of time corresponding to the Smat (float vector of length = NoofPSDs) 1.13. Cross-spectrum, Transfer and coherence functions [Stat, TRNvec, xVec, COHvec] = Tran(Vec1, Vec2, Offset, Length, FrameSize, NoofFrames, SampRate, FFTSize, Winow, overlap) Calculates the transfer function and the coherence between two time histories (raw data or other 2 vectors). Received Parameters: 13 US 7,027,953 B2 55 56 0 Vecl - time history of _- input - vector (raw data vector or another vector) beginning at Offset of length Length points: vector of integers of length 256-500000 samples. a Vec2 — time history of - output - vector (raw data vector or another vector) beginning at Offset of length Length points: vector of integers of length 256-500000 samples Offset - start point in the raw data matrix (long integer) Length - length of vector Vec (long integer) FrameSize - frame size for calculation (integer) NootFrames - no of frames to average (integer) SampRate - Sampling rate of the raw data (float) FFTSize - frame size of the fit output vector*2 - enables zero padding (integer) Window - code for window type to apply on Vecl and Vec2 (integer) 0 = "Rectangular" 1 = "Hanning" (default) 2 = "Hanuning" 3 = "Blackman" - overlap - overlap percentage (float in the range 0-100 (0 default» Returned Values: 0 Stat - status / error code (integer) n TRNvec - transfer function vector (complex vector of length = FFTSize/Z) o xVec - vector of frequencies/orders corresponding to the Tvec (float vector of length = FFTSize/Z) . 0 COHvec — coherence vector (float vector of length = FFTSize/Z) Algorithm: Gt) (f) Gxx (f) M; Gr. (f) * ny (f) 0” (f) - Auto—spectral density function. 0,, (f) — Cross-spectral density function. Transfer function defined as: Coherence fimction defined as: 1. 14. WignerViI/e This function calculates Smoothed Wigner-Ville time-frequency distribution. [Stat,tfr,t,t]=WignerViIle(pathName,mode,channel,Tstart,Tend,Fstart,Fdelta,dF); Received parameters : 0 pathName - experiment path. (string) ' 0 mode - mode name. (string) 0 channel - channel number. (integer) 0 Tstart - start time. (float) o Tend - end time. (float) 14 US 7,027,953 B2 57 58 3? H0 in: ww- -. m .J - .r ..r' “.1‘l'Luzé-I;§ll't'f'rl.l.2‘§f o Fstart - start frequency. (float) o Fdelta - range of frequency ( Fstart : Fstart+Fdelta ). (float) o dF - frequency resolution. (float) Returned values : o Stat - error/success code. 0 tfr - Wigner-Ville matrix. a t - time vector. 0 f — fi-equency vector. Algorithm: [XmIZ : .iix(").x(u)€_mm_v’dudv x‘ - complex conjugate of x. 1. 15. Filtering function [Stat, Fvec]=filtering(Vec, F iltType, SampRate, MinFreq, MaxFreq, Slope, PdB, PhaseType) This function enables filtering of time series Types of filter: LP, HP, BP, BS. The filter is defined by its transfer function. Vec - time history vector (raw data vector or another vector describing data in the time domain) beginning at Offset of length Length points: vector of integers of length 256-5 00000 samples 0 FiltType - Code for filter type: 0 - filter no needed,1 - LP, 2 — BP, 3 - HP, 4 — BS. SampRate — Sampling rate of the raw data (float) MinFreq - ignored for LP and HP filters and the start freq. for BP and BS filters MaxFreq - the cutofic freq. for LP and HP filters and the end freq. for BP and BS filters Slope - filters slope in [dB/Oct] (integer) PdB - Level at cutoff frequencies [dB} (integer) PhaseType — filter phase distortion code: with/out (integer) ' O-filtfilt 1-filter Returned values: 0 Stat - status / error code a F vec - filtered vector time history(float vector of length equal to Length) Algorithm: 1. Parameters validity check (depending on filter type) 15 US 7,027,953 B2 59 60 '4} u”)! “‘2'! H B M II “a "" ---- .iL. LII .3: . .“'B"'1r'.a" V" .- 32“: fl 2. Prepare parameters for Butterworth order determination (function ‘PrepFilt’ calculates the order and coefficients (a,b) of the requested Butterworth filter) Activate filter/filtfilt filters. 16 '23: M: a": 11.- nZJI 6.“)! 11..» 9' .1.» 62 “SI” "‘ Jw¢73 -N ["16 all. u...“ m US 7,027,953 B2 61 T< 9an _. 99m - mwm. mumi wEEPP $391 33 — méh - mmm. 253 9.5.3. 939m 5; I II II I ll m 0.9m - mmm . mumfi SwmmtEou Sammi 2;: m muflm - mmm. mum—n .6wa. Eco oSmEn. 5.2.. II II II II II ,_ II. 8822 gfi HE l_ ‘— I III I l I m I I NWFC") IE- 0 o C") I I ll 888m! IIEII ll 808.3! IEEII 2588! I I NOD m I'll g ll '3! n $.8m - mmm. mama 539.88 93th 2:: m $.me - mwm. mum—n 539.258 0.53.1 2...: m w.m.m - mmm. muma homwetEoU 932m. 5:: v w.m«m - wmm. oumfi .omm2.E8 952.“. z ._I I m ml I II l 889% ; IEII 889E ! I III. II lléll I 8082! N lugll lg! llall I I I I I I I MNMNWNMN m mtflm . mwm. oumE 33$.E8 8:981 z: N 33m - mmm. mumE .cmmmfiEoo 9505 5:1 w m.mwwm - mm“. wUNE howmmgoEOO mEDwQ—n‘ £-_I m 99m - mmm. mumE .ommw...E8 958.5 26 l l l I I l I I I l I I I I I I v w.m.m - wmm. £55 .032. E8 9521 25.. l I I I I l I I m m !E| m ! l '5! l l l nnwnmm F I '5! l g m m.£m - mmm. moma .omm2.E8 95me 33 l N mmflm - «mm. 253 .ommE.Eoo 2:35 33 3.2m - wmm. 822 cm“. I l ._ I H E II I In! I I llé I Ila I I I 7 cm: '5 UN: IEII 5! IE! lg; lg! UcunwEm E n 6 03.59“. _. l o l o l l 808.8 g a u o o I N N N N N N N N N N N N N N _, _. _. w v m N Frmmmmn ! l I l l l l l ! In :95 2391 25.. I r wuoo £235 z< E IRE I II I! ll IFI ll Fl ll rI II II I! I I! I II I ll I ll I II I ll lpl II I! F! I II I II F . r ll a. an n. 64 "'4' J‘ flflflW‘ J .J “11“?!” m «n W")! ...n. ILMu ...:.lx US 7,027,953 B2 63 N-< $an 89 55° . m % 5E8 Igllllll 883: Iwfi lllllul 303'me llllll 83:2 IE, 98 . w 3. .553 glllllul 883 Ina ll'll!!!all . IEIIIIII 8%.? IEE gllllll 33% IE Illl!!!=_|| IEIIIIII 3:38 IE Ifillllll fide IE Ellllllélfl 'MEIIIIII 983 IE llll!!!§l| IEIIHIIIEIQ gllllllélfi gllllllég gllllll £82 £2 lllllfi! 9;. ll IEIIIIIIE! IEIIIIII «83¢ IE llll!!!all 82 .28 . m g 5.8% gllllllglfiu 5.» =3 . N x .553. IEIIIIII 88% one 89 .85 . N k .Emwm .wmllllllé! “:8 . r * .Emwm IEIIIIII :83 ! lllll!!!all 89 .98 - F u .558 IEIIIIIIE! =3 =2 - . 3, 5.5% Igllullllélfia 82 3E - . % SE3 gllllllé! llllllmllall v 9% . m8. 255 23:: 2:35 23 lllllll 8832 IE llllllluléflll m 95 - am. 8% 2.5:: 2:85 23 lllllll 088de ! lllllllgill N 32w - m3. 3% 85:; 2391 :3 Ill—IIIIE! US 7,027,953 B2 65 66 2!! llllllllllllllllllllllllI!I!llllllllllllllllllllllllI'llU3lllllllllll! O O O O C C O O O O O -ear mesh Description PTO bearim inner race near mesh -ear mesh earshafl accesso 1 s 0 ur 4 s-ur oear mesh - axis 8 bevel near mesh 0 ube oumn shaft and HMU shaft fuel boost cum. se narator blower shaft .2 U) a: E E a: . I— :I . {n N 0 V) é’ U N accesso 3 s n ur accesso Bearint # 5 - caoe alternator shaft ‘M - 4 Harmonles CHAN Weigh RPM Code '5 l 5 m m mm .9 o 2 . V1 2 .. . 8 cu m - '0 19 .- .— *— .—.— .— ifi ' -° » 2 2 2 22 2 o o o o o o 0 0000000 o o o o o o r9: mwmmmmm In In In I!) n N 0000 m mrchomrxv v v V: v m g coco co ‘rranvmn! N. or n! or Q m ' ‘QVNN‘QNm co co co :0 N . o oooow- v v V v- N In O O O O 0 “2°..- °°.oo' . CF67 OF) C o o- u c .- le1abm — — gg 2 Pointer Codes Ie1acm4 |e1acm5 e1bs 1acm1 I Ie1acm2 Table A-3 2'5) “It 11..“ u .1:- "9? M M. at?“ «0 ....LL. 68 .nl" “ad-w .4' 49in» 3: “r "11" IF?‘ ..;n.. ILnD ... .— .5. US 7,027,953 B2 67 v-< $an 98 game ; §.§_E!E!!!!ll 883 _ 5% 5.3.35 ; é§__glfil!!!ll 882 _ fies, 38550253 5 2:823; la!!!!!l 8%? _ :25. 'llll!l!§|| 89553235 : .Emmfiglfilllgll 83.3; as .2289 .EgeanlfillllEII 8&3 . :3 .2989 atme E 309.23 .2229 .cfis__slfill_!!_!| 883% ,ll'll!llflm_|| 82 .25 555%. .553 __2 IE!!!IE!I 025.2 IE 28 am: 5. .3 5E2 5:2 la!!!!!l agno :5 a? E. . a 5E2 .22 la!!!!ll 053 E .82 .28 a? E. .3 .553 5:2 glll!ll 888 g lllllgllmfill 82 5E a? E. .3 gas 5:9 glll!llélmg mac $352229 .Em35__elfilll!ll anogé =5 535222? .=__§a__e l3!!!!!l OEEIE-EE .89 .28 .592 55%? Egg 5:2 lull!!ll 83.0.2 Inn-EV 'llll!llflallfl 89 as: .525; .2225. £58 5:9 l5!!!!!l 8m 5. ; lam-E 28 5:25 0E ll!!!!l 838; 5; 2:03 OE Ill—lgllé £6. 38:23 .=__§oE'II_!!!| as; 352 llll!llwfi_l _ US 7,027,953 B2 69 70 u u n u w... . :, ---n-'-1r':’t’I Jr?! 53 M? In .meflm IF; wngsdaa v- v- v- v- v- I— 1— ‘— ‘— :— ed0|eAua > SBBJOAV aseqd 8 adolonua Proccessing Type ('1' -relevant) JepJo C 0) (D U U E S - _ 0.) G) C C s g # 3 — inner race # 3.1 - outer race Descri tion as o S L- a) C .E . ‘_ m 4:: . .E h m a) m Bearin- # 3.1 - ball 5- Bearino # 2 - ball s-in Low Presure Shaft spueqapls SJa)u!0d ._DO U . 1—1—V— ‘— 929 2 muss aagnaa bearin . 1 IL... Table B-1 US 7,027,953 B2 71 72 "H I“)! “7;” AL. L." ,‘m . “WW 2-H! Hp '39.» 2;; .le 1M «.1- "2:. pi II . Fa!" .J" In . tn 0 Io ‘— ‘— ‘— ‘— ‘— ‘— ‘— ‘— ‘— C 1 II II II II aBeJaAv aseqd 1- !- 1- ‘- I— ‘— ‘— ‘— ‘— ‘— adolanug ‘— r- ‘— r~ ‘— r- 1- ‘— r- \— I— r- ‘— 1- JapJo m Proccessing Type ('1‘ -relevant) a) 0 Egg oc§ 5- 5 5 m ._._ (ch—L . . .. c c: a- E ta~ 9 9 9 0 ma) __0J _, q.)_m % m a) m -- .CE _""‘ C_“ C C C 4d «1:: I (U: .C 03.; .00 $395.00 :0 8, 8, 8., I: _a)I II 1-— Ill ‘1’ g 5::- Vrv 'agmmm 5 23 Eng) on 3:: an: gwuuu u; t :8: n L 59 E «1 (“gm “E 2% a 2 3-2 .Cm zwm 3 __ _q)__ 0') ma] — —-o—l—— .-a.) .._._a.) o m man: In) Ian: _I .o. 20.0 spuerwms a) O c ’ - .DO U ' IDLO (D 22 n apoo wawgod Ib5c 390W muss aogAaa beann- generator Table B-2 US 7,027,953 B2 73 74 n» W '3}: II II I”) ""'-W "ng -“ “’3'! TH l'"b M ‘33“ mm. ILJI 'lr"1fl" ,.:-' Jr as "an n... “:3: awn 11.4: 4:: puncJB 01398 ‘ VHd 1- ‘— s- r' I— \— ‘— I— \— wnJISdao F ‘— adolaAua I aSBlld 1- ‘— P 1- I— I— 1- v- ‘— adOIBAu a --I-“ ---i Proccessing Type ('1' -re|evant) Description ball bearing generator inner race outer race” spueqaprs epow emued aogAaa - enerator ball bearing Table B-3 US 7,027,953 B2 75 76 . . n *- r v-w « - E" ..:n..'n..n ..'::. ‘ "If" "my? 1' m 11:: .5: IKE? “III I “EHIIIIIEIIEEII “Illllli Mlllliilli m-llii-Ili I mullflll “II-Ilia Ell-IIII-IIII puncJB 61398 wnnsdao P ‘— eddIaAua v. .— 1— F V‘ ‘- " ‘— ‘— -umnsd93 I v- 7- ‘— afie‘al‘v ‘— F F ‘— ‘_ ‘— aseqd - ' ’ " F " r F F F adolanua ‘_ ‘_ ‘- F v— ‘— \—- ‘— ‘— ‘— :- Proccessing Type ('1' -relevant) clutch shaft Descriptlon generator shaft accessory wheel shaft accesso 4 near mesh h drauiic uum. shaft accessory 3 shaft epoo spunmms apoo smwwd E E W mm 0 U .— 8 m m «5.9- a) a) a: (Do accessory4 o ear le1acm4 N (’3 Z‘ POW g 8 sinned g 3 . aognau 8 8 g m N ' Table B-4 US 7,027,953 B2 77 78 4|mw%uwmwwmw «I AM. "1‘. II I} I u "‘3' ""9' . 7 «1r r' In -11. In; “:3 Lulu: 1:; «In ILJI .22» ""1T‘.-F a VHd ‘- ‘— wnnsuao ‘- r ‘— F ‘— adolenug *- 1— ‘— ‘— ‘— \— aseqd F 1- !— I— ‘— iillliilllllllli- milllllli- Proccessing Type ('1' -relevant) near mesh .ear mesh Description Low Presure Shaft oenerator shaft hydraulic nump shaft accessory 3 shaft accesso 6 sear mesh accesso 5 a1sr apow alnued amnea L (U Q) o accessory5 Table B-5 US 7,027,953 B2 79 80 ‘un - . ‘1? ‘“.' “ " “.1 m1; Hall r3"? .. 3.1:"..‘5 will}??? umusdeo odoIaAua 3': C m > 2 a) I— I ;_ La a: E. p. U) .E 8 d) U u E A Description Low Presure compressor blade nass - Staae 2 ape-3 spueqapls apoo smaugod apow amued cognac Fan blade pass - Stan e1 Presure compressor compressor Table B-6 82 H»!H""'""§P , ‘1r“-1r .J“ Hull .u'l'u «a 1m. -.: US 7,027,953 B2 81 Wm 22m... $3 263 5mm9nE8 289m :9: wmma mums 8392.8 239.”. - I F .owmwano m5me punOJB Lunnsdeg VHd l-I 3:553... .5 2;... @6388; 553.530 spuaqepm rm #“b 2‘ -..‘u 11..." L» -n 1;" "' 84 US 7,027,953 B2 83 m-m 29m... mwmd mugs memano 239m :9: mmma mvmfi .88an0 9:85 :9: ‘— mmma 255 5823.8 238.5 :9: — _ _ E u A a m. d a O 03 m mm 9 $8 1' 1' J Jo n "d m we ma Bu. 12 .m“ a $2929.- .5 on»... mEmmouuoE spueqams 5'3» M5 53:; .- .331 “am w- ».u m .u v.-n.. ll: 86 Ii 3 JI fl “’3‘” "'11"er .ul" ".1 ~41 r-u u-flu ll—l US 7,027,953 B2 85 Qm 2an 8mm 8% 5mmmaE8 2:85 :9: 0 0.9m . mwm. mama o «:me kommmano Hamel :9: $3 233 Emmmano 2:35 38 mums Lemmanou 959m :9: US 7,027,953 B2 87 88 mu m. w .0“ 1L.“ .3. II D I! I "‘3' "“W J"- :l “r "11".! _lll ' VHd tunnsdeg adolanua 1— ‘— P _l-I-I afiuanv T7 Fl I aseud _I_l_l Proccessing Type ('1' -relevant) Description L o U) ‘0 cu L. o. E o O ‘2 3 U} ‘2 D. .C .9 I I— o to ID ‘2 O) o. a: E o 6-4 U (D OJ l L m S m 9 Q- 0 .c '0 .9 L“ I a do a: O a: +1 0') l w w m . cu 'U L“ .0 High Presure Turbine blade nass — Sta-e1 690:) SPUEQBP!S apow amued» ODIAOO blade pass blade pass Stan e ‘1 compressor 4» 113 “m w. m m m _(L. n... .212 IL)! 11.11! 1LT. Table 8-10 90 US 7,027,953 B2 89 F Tm 2an h “'“8 II..." It; warm: u" MIDI 1L.” 3.55% Ezwmi m). ‘5 n -11“! .. q" ":r .r' ""n- .m -n )1 u T 2:23, 3391 "a. mu "D 3"» a * .ulJ.. ILZII , 059.3 9335 9.55% 928.1 mmxwflr: 91 US 7,027,953 B2 m— mum— m- NPenineshaft —aNGenineshan Table 0' US 7,027,953 B2 93 94 What is claimed is: ing index functions, the pointer locations and index 1. A health maintenance system for a mechanical system, functions associated with known faults of the mechani- comprising: cal system; and A Vibration sensor, the Vibration sensor acquiring Vibra- comparing the feature data with feature data associated tional data relating to the mechanical system; 5 with known system faults to provide a health indication A data processing circuit, the data processing circuit for the meehahical system. including a transformation module transforming the 13. A method for diagHOSing the health of a mechanical data acquired by the Vibrational sensor to signature data system, the methOd COmPI‘iSing the steps of: in a plurality of domains, the signature data associated receiVing Vibrational data from at least one sensor moni- with at least one predetermined pointer location corre- 10 toring the mechanical system; Spending to known fauns 0f the meehanieal System; processing the Vibrational data to pr0Vide features asso- A diagnosis circuit, the diagnosis circuit extracting fea- ciated with said mechanical system; tures from the signature data by reference to said comparing the pr0Vided features with features associated pointer location associated each signature data, the data with of known fault conditions; and diagnosis module employing at least one index func- t5 determining whether a fault condition should be indicated tion to pr0Vide features corresponding to each pointer for said mechanical system based on said comparing. location and signature data; 14. A computer implemented health diagnostic system for A decision circuit, the decision circuit determining a a meehameal SyStems eemPriSing health status for the mechanical System by reference to at least one Vibration sensor positioned to collect Vibration the features provided by the diagnosis Circuit. 20 data of at least one component of said mechanical 2. The system of claim 1 wherein said mechanical system SyStemé is an aircraft engine. a data processing circuit, the data processing circuit 3. The system of claim 1 further comprising at least one ihehlding a proceSSing unit WhiCh transforms the Vibra- gas path sensor. tion data into signatures in a plurality of domains, each 4. Ahealth prognosis method for a mechanical system, the 25 signature data in each domain is associated with at least method comprising the steps of: one pointer, said pointers related to at least one known monitoring at least one operational parameter of said fan“ condition 0f the meehanieal SyStemé mechanical system; a diagnostic circuit, the diagnostic circuit employs an detecting an anomaly in said Operational parameters; index function of each pointer location to extract fea- recording at least Vibrational data in response to detecting 30 mm data from 631011 Slgnamrei and said anomaly; a decision circuit, the decision circuit deriVing a health processing said Vibrational data to pr0Vide features asso- Status from sald feature data eXtraCted by the (hat-5110515 circuit. 15. The system of claim 14, wherein the Vibration sensor 35 is further positioned to collect Vibrational data from the exterior of enclosure case of said at least one component of the mechanical system. 16. The system of claim 14, wherein said mechanical system is a rotating machine. 40 17. The system of claim 16, wherein said mechanical system is an aircraft engine. 18. The system of claim 14, wherein said diagnostic circuit further aggregates features of the same pointer location, across a plurality of domains. 45 19. The system of claim 14, further comprising at least one non-Vibration sensor positioned to collect data associ- ated with the mechanical system. 20. The system of claim 14, wherein said at least one non-Vibration sensor is a gas path data sensor for collecting 50 mechanical system gas path data. 21. The system of claim 19, further comprising a storage module adapted to record said Vibration and said gas path data. 22. The system of claim 14, further comprising an opera- 55 tor display adapted to communicate an indication of mechanical system health. ciated with said mechanical system; trending said features by a prognosis module to pr0Vide an expected future feature data; and analyzing said future feature data to determine an expected effect on said mechanical system. 5. The method of claim 4 further comprising the step of downloading to at least one communication ground station the data acquired in response to detecting the presence of said anomaly. 6. The method of claim 5 wherein said downloading step Via wireless link to said ground stations. 7. The method of claim 5 wherein said downloading step is performed periodically during operation of said mechani- cal system. 8. The method of claim 5 wherein said downloading step further comprises the step of downloading said anomalies to an off-line data extraction application. 9. The method of claim 4 wherein said recording step includes the recording of an anomaly code indicatiVe of the detected anomaly, and historical data releVant to said anomaly. 10. The method of claim 4 wherein said trending step is triggered by one or more predefined states. 11' The methOd Of Clalm 4 further compnsmg the Steps Of: 23. The system of claim 19, wherein the decision circuit Compulmg a? least one hfe H‘sage mdlcatorsg and is adapted to receiVe said gas path data for deriVing said recording sa1d llfe usage 1nd1cators. health status. 12. A method for pr0Viding a health indication for a 60 . . . 24. The system of claim 14, wherein the diagnostic index mechamcal system, compns1ng: function comprises a function for comparing a baseline receiVing Vibrational data from a data eelleetien unit; signature extracted during normal system operation with a processing the Vibrational data to pr0Vide at least one signature provided by the data processing circuit. Wide-band frequency-domain Signature and at least one 25. The system of claim 14 wherein said diagnostic circuit Other Signature in another domain; 65 further assigns relatiVe weight to each feature data item extracting feature data from said signatures by reference associated with said known fault condition of the system by to pointer locations for the signatures and correspond- a corresponding pointer location. US 7,027,953 B2 95 26. A health prognostic method for a mechanical system, the method comprising: detecting a deviation of operational data, including vibra- tional data, from a predetermined normal; recording at least Vibration data in response to detecting said anomaly; transforming the Vibration data into signatures in a plu- rality of domains; extracting features from said signatures to provide fea- tures that are indicative of a health status of a compo- nent of said mechanical system; trending said extracted features to provide expected future feature data; and analyzing said future feature data to provide a prognosis of an expected parameter state for said mechanical system. 27. The method of claim 26 further comprising downloading, to at least one communication ground station, the vibration data acquired in response to detecting the presence of said deviation. 28. The method of claim 27 wherein said downloading step is via wireless link to said ground stations. 29. The method of claim 27 wherein said downloading step is performed periodically during operation of said mechanical system. 30. The method of claim 27 wherein said downloading step is performed subsequent to operation of said mechani- cal system. 31. The method of claim 27 wherein said downloading step further comprises the step of recording said parameter deviation data by an off-line data extraction application. 32. The method of claim 26 wherein said recording step includes the recording of a parameter deviation code indica- tive of the detected parameter deviation, along with record- ing data relevant to said parameter deviation. 33. The method of claim 26 wherein said recording step is triggered by one or more predefined system states. 34. The method of claim 26 wherein said trending step employs polynomial fitting to predict said expected param- eter state for the mechanical system. 35. The method of claim 26 wherein said trending step employs adaptive reasoning parametric models to predict said expected parameter state for the mechanical system. 36. The method of claim 26 further comprising the steps of: computing at least one life usage indicator; and recording said life usage indicator. 37. The method of claim 26, wherein said mechanical system includes at least one inter-shaft bearing, whereby the method detects deterioration of said inter-shaft bearing. 38. A method for providing a health indication for a mechanical system, comprising: 10 15 20 25 30 35 40 45 50 55 96 collecting vibration data by a data collection unit; processing the vibration data to provide at least one wide-band frequency-domain signature and at least one other signature in another domain; extracting feature data from said signatures by reference to at least one pointer location for the signatures and by employing a corresponding index function, the pointer location and index function associated with known faults of the mechanical system; and comparing the feature data with feature data associated with known system faults to provide a health indication for the mechanical system. 39. A method for diagnosing the health of a mechanical system, comprising: collecting vibration data by employing at least one vibra- tion sensor; deriving vibration signatures in multiple domains by transforming the vibration data to multiple domains; extracting data from each signature by reference to pre- determined pointer data; employing an index function to extract feature data from the vibration signatures in the various domains by comparing said extracted data to baseline data associ- ated with the index function; aggregating feature data associated with a known fault; comparing said feature data to a predetermined threshold to diagnose the health of the mechanical system. 40. The method of claim 39, wherein said pointer data defines at least one domain region applicable to a known system fault. 41. The method of claim 40, wherein the pointer data defined region is compared to a baseline region by using diagnostic indices. 42. The method of claim 39, wherein each index provides feature data by reference to deviation of the pointer defined region from the same region of baseline data. 43. The method of claim 39, further comprising aggre- gating the feature data to arrive at a health diagnosis for the mechanical system. 44. The method of claim 43, wherein said aggregating further includes combining the extracted features with fea- tures from gas path data. 45. The method of claim 39, further comprising predicting failures of the mechanical system by comparing the extracted features to historical data relating to known faults. 46. The method of claim 39, wherein said domains are selected from the group consisting of time, order, quefrency, time-frequency response, amplitude, parameters, rotations per second, RPS-order, cycles, envelope, phase average, orders of envelope, cepstrum of envelope, and background spectra in the frequency domain. 47. The method of claim 39, wherein said signature analysis is wideband from 0 to 5 KHZ. * * * * * ...
View Full Document

Page1 / 58

7027953_Method_and_system_for_diagnostic - USOO7027953B2...

This preview shows document pages 1 - 58. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online