ANN_AGING - Prediction of thermal ageing in transformer oil...

Info iconThis preview shows pages 1–2. Sign up to view the full content.

View Full Document Right Arrow Icon
Prediction of thermal ageing in transformer oil and high voltage PVC cables using artificial neural nemo r ks L. Mokhnacke, A. Boubakeur and A. Feliachi Abstract: The prediction of the thermal ageing of transformer oil and of high voltage PVC cables using artificial neural networks (ANNs) is discussed. This prediction provides an altemative to costly laboratory experiments. To validate the proposed technique, predicted and laboratory results are compared. The ANN used is a radial basis function Gaussian network trained by a random optimisation method 1 Introduction Currently, the electric utilities of many countries are being restructured, and as a consequence cost savings are becoming increasingly important. One way of reducing expenditure is through proper monitoring and maintenance of power equipment, such as large transformers whose major cause of failure is thermal ageing of their insulation materials [I] (oil and paper). The dielectric and mechanical properties of high voltage (HV) and medium voltage (MV) cables made up of XLPE cross-linked polyethylene are significantly affected by thermal ageing [2]. The same is true for PVC cables [3, 41. Neural networks will be used to predict the thermal ageing behaviour of both the transformer oil and the PVC used in cables. Experiments on the thermal ageing of PVC insulation have been performed using the test facility of CABEL, an Algerian cable company. The test-probe experiments were carried out at different temperatures ranging from 80°C to 140°C with a maximum ageing duration time of 5000h. The tests consist in ageing the material and measuring dielectric and mechanical properties at regular time intervals [3]. We have performed experi- ments on the thermal ageing of BORAK22 transformer oil, which is used by SONELGAZ, the Algerian National Electric and Gas Company. These experiments are costly and time consuming. Our current aim is to reduce the ageing experiment time and predict the thermal ageing stress for longer times using artificial neural networks (ANNs). Having a database of the full thermal ageing interval, we intend to train our net to predict insulation characteristics at longer intervals. We hope to train the net within an interval that is less than the one used in the experiments, whilst being able to predict insulation 0 IEE, 2003 IEE Prucurdings online no. 20030268 dot IO. IM9,ipsmt:2W30268 Paper 6ril meiued 17th May 2002 and in rwid form 10th Decemkr 2002 L. Mokhnacke is \\lth the lnstitu~ d'Elctrotechnique. Uoirersilb de Balm Batna, Algeria A. Boubakeur is with the Ecole Niitionele Polytechnique. B.P. 182 El-Harmch, Algicn. Algeria Feliachi is with the Lane Department of Computer Science & Elec1"cal Engineeing. West Virginia University, P.O. Box 6109. Morgantoan. WV 26506-61O9. USA IEE Proc-Sci.
Background image of page 1

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

View Full DocumentRight Arrow Icon
Image of page 2
This is the end of the preview. Sign up to access the rest of the document.

This note was uploaded on 06/08/2011 for the course ELECTRICAL 124 taught by Professor Ghjk during the Spring '11 term at Institute of Technology.

Page1 / 6

ANN_AGING - Prediction of thermal ageing in transformer oil...

This preview shows document pages 1 - 2. Sign up to view the full document.

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