基于BP神经网络的高压断路器故障诊断

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

为诊断与分析高压断路器故障,本文提出了基于BP神经网络的高压断路器故障诊断方法。该方法利用高压断路器典型分合闸线圈电流-时间曲线,能反映其机械故障状况的特点,将仿真输出数据与故障编码比较获得诊断结果。该方法只需一组完整的故障数据作为网络的训练和测试输入,就能够诊断出高压断路器操动机构是否出现异常情况,以及确定出现故障的类型。本文以MATLAB2014b为试验平台,用实际数据作为训练样本和测试样本进行仿真分析,其输出结果与期望输出一致,验证了该方法是一种有效的高压断路器故障诊断方法,具有广阔的应用前景。

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.