Turn to Turn Fault Diagnosis for Induction Machines Based on Wavelet Transformation and BP Neural Network

Najafi A., Iskender I., Farhadi P., Najafi B.

International Aegean Conference on Electrical Machines and Power Electronics / Electromotion Joint Conference, İstanbul, Turkey, 8 - 10 September 2011, pp.294-297 identifier

  • Publication Type: Conference Paper / Full Text
  • City: İstanbul
  • Country: Turkey
  • Page Numbers: pp.294-297
  • Middle East Technical University Affiliated: No

Abstract

Based upon Wavelet Transformation analysis and BP neural network, a method for the fault diagnosis of stator winding is proposed in this paper. Firstly wavelet transformation was used to decompose vibration time signal of stator to extract the characteristic values - wavelet transformation energy, and features were input in to the BP NN. After training the BP NN could be used to identify the stator winding fault (Turn to Turn fault) patterns. Three typical turn to turn faults as 10 turn, 20 turn and 35 turn were studied. The result showed that the method of BP NN with wavelet transformation could not only detect the exiting of the fault in stator winding, but also effectively identify the fault patterns.