Traction Locomotive Incipient Fault Analysis using Wavelet based Particle Swarm Optimization

The present work aspires to pursue a qualitative research on the loosening of brushes in Indian induction motor locomotives, observing the traction transformer input current. At the initial step, the locomotive has been built in programming and the current input to the transformer has been documented at normal condition. Brushes have been gradually loosened and the transformer inputs have again been recorded. Both the currents have been assessed using a novel Wavelet Transform (WT) coefficient statistical analysis based Particle Swarm Optimization (PSO) Technique. From the computed Skewness, Kurtosis and RMS values, global bests have been disintegrated and optimization has been done in PSO plane to assess different conditions in the engine associated with an incipient brush fault. Here multi-objective PSO (MOPSO) has been used for the assessment and decision making. With the outcome of the refinement, a credo has been developed at the end to segregate the faulty traction engine from the healthy one. The technique used here has been validated in offline mode with satisfactory outcome.