AE02.pdf

System which will identify the class to which each

This preview shows page 24 - 26 out of 82 pages.

system, which will identify the class to which each unknown signal belongs. Successful identification of input signals is a good indication of a properly trained system. Signal Classification The data used to train a classifier correspond to a group of points in feature space. If two features are used for a system with three source classes, then the feature space might appear as shown in Fig. 8. The classification mechanism must distinguish among the classes. One simple way to distinguish the classes is to place planes between the classes (hyperplanes in a multidimensional space with many features). The classifier is a linear combination of feature elements that defines a hyperplane to separate one class of signals from another in the feature space. This is called a linear discriminant function classifier. When the classes are distributed in a more complex arrangement, more sophisticated classifiers are required. An example would be the K nearest neighbor classifier that considers every member in the training set as a representation point. It determines the distance of an unknown signal from every pattern in the training set and it thus finds the K nearest patterns to the unknown signal. The unknown is then assigned to the class in which the majority of the K nearest neighbors belongs. In other cases, the pattern recognition process is modeled statistically and statistical discriminant functions are derived. Signal Treatment and Classification System A classification system is shown in Fig. 9. Acoustic emission source characterization and recognition begins with signal treatment. The acoustic emission signals from different sources are first fed into a treatment system that computes the necessary waveform features of each signal. The identity of the signal is tagged to each of the feature vectors. These vectors are then divided into two data sets, one for training the classifier and the other to test its performance. Generally, a normally distributed random variable is used for assigning the signal to the two data sets and this ensures that each set has the same a priori probability. Once the classifier is trained, it is tested by supplying a signal feature vector with a known identity and then 49 Fundamentals of Acoustic Emission Testing F IGURE 8. Feature space for two features used for a system with three source classes: (a) linearly separable; (b) nonlinearly separable; (c) piecewise linear separation of two regions. (a) (b) (c) + + F IGURE 9. A classification system for treatment of acoustic emission signals. Batch control file Batch feature extraction or Preprocessor Extracted feature data Training sample Control Testing sample Feature selection and decision logic Trained information Classifier Trained information Decision library Classification output Signal data
Image of page 24

Subscribe to view the full document.

comparing the known information to the classification output. The training information and the trained classifier setup can be stored in a decision library that can be used to characterize and classify the acoustic emission sources.
Image of page 25
Image of page 26
  • Fall '19
  • Nondestructive testing, Acoustic Emission, Acoustic Emission Testing

{[ snackBarMessage ]}

What students are saying

  • Left Quote Icon

    As a current student on this bumpy collegiate pathway, I stumbled upon Course Hero, where I can find study resources for nearly all my courses, get online help from tutors 24/7, and even share my old projects, papers, and lecture notes with other students.

    Student Picture

    Kiran Temple University Fox School of Business ‘17, Course Hero Intern

  • Left Quote Icon

    I cannot even describe how much Course Hero helped me this summer. It’s truly become something I can always rely on and help me. In the end, I was not only able to survive summer classes, but I was able to thrive thanks to Course Hero.

    Student Picture

    Dana University of Pennsylvania ‘17, Course Hero Intern

  • Left Quote Icon

    The ability to access any university’s resources through Course Hero proved invaluable in my case. I was behind on Tulane coursework and actually used UCLA’s materials to help me move forward and get everything together on time.

    Student Picture

    Jill Tulane University ‘16, Course Hero Intern