A Novel Life Prediction Method for Railway Safety Relays Using Degradation Parameters

In this paper, a novel life prediction method is proposed for analyzing and assessing the reliability of railway safety relays. In the life tests of railway safety relays, the authors blend the principal component analysis (PCA) with Mahalanobis distance to extract the key degradation features from some parameters, including the relay contact resistance, closing time, releasing time, super-path time, arc time, and bounce time. The safety relays are classified into three types according to their different failure mechanisms. The optimal identification parameters are regarded as the prediction variables using the Fisher discrimination criterion. Then, adopting the back propagation (BP) Neural Network to train the prediction model so as to predict the change trend of the parameters. Finally, the prediction model is validated against the real life data of failure relays, using wavelet transform for eliminating noise. Simulation results demonstrate that the proposed method has high accuracy of life prediction for railway relays and can be easily implemented in real engineering applications.

Language

  • English

Media Info

Subject/Index Terms

Filing Info

  • Accession Number: 01678525
  • Record Type: Publication
  • Files: TRIS
  • Created Date: Aug 2 2018 2:56PM