Novel BP Neural Network Model for Fault Diagnosis of High Voltage Circuit Breaker

A novel BP neural network model for fault diagnosis of high voltage vacuum circuit breaker (VCB) was developed.The influence of a variety of possible mechanical faults on the characteristics of the time evolution of a typical switching coil current, divided into five distinctive stages, was mathematically modeled, experimentally evaluated and numerically simulated with MATLAB2014 b as the platform to diagnose a specific failure of VCB.A complete set of the measured abnormal switching-coil current evolution were acquired as the training, testing and simulation samples.The preliminary results show that a specific VCB fault has a major impact on the time-dependent switching coil current.The predicted, simulated and measured results were found to be in good agreement.We suggest that the newly-developed BP neural network model be capable of effectively diagnosing fault of high voltage VCB.