基于特征提取的有源电力滤波器故障诊断方法.pdfVIP

学兔兔 第52卷 第15期 电测与仪表 V01．52 NO．15 2015年 8月10日 Electrical Measurement＆ Instrumentation Aug．10，2015 基于特征提取的有源电力滤波器故障诊断方法术 马立新，吴兴锋，穆清伦 (上海理工大学光电信息与计算机工程学院，上海200093) 摘要：针对有源电力滤波器APF(Active Power Filter)的IGBT功率管易发生故障的问题，提出了基于故障特征 提取的有源电力滤波器故障诊断方法。构建了APF故障仿真模型和基于小波包分析的故障特征提取方法，仿 真分析了APF网侧电流波形，并运用小波包分析对IGBT故障时的网侧电流波形进行处理，提取了IGBT故障 特征向量，最后运用神经网络对特征向量的分类来实现对APF的故障诊断。在APF故障诊断系统上进行测 试，验证了该诊断方法的有效性和可行性。 关键词：有源电力滤波器；特征提取；故障诊断；小波包；神经网络 中图分类号：TM714 文献标识码：B 文章编号：1001—1390(2015)15—0124—05 Fault diagnosis method for active power filter based on feature extraction Ma Lixin，Wu Xingfeng，Mu Qinglun (School of Optical—Electrical and Computer Engineering，University of Shanghai _厂0r Science and Technology，Shanghai 200093，China) Abstract：For the sake of easily damaged characteristic of IGBT，a method to diagnose the fault of APF was proposed based on fault feature extraction．The model of APF fault simulation and the method of the feature extraction based on the wavelet packet analysis were established，and the waveforms of grid side current were simulated，processed the waveforms of grid side current of IGBT open circuit fault by wavelet packet analysis，and extracted the fault features vector．Finally，the classification function of the neural network was applied for the APF fault diagnosis．The effec— tiveness and feasibility of the diagnosis method are validated by test results of APF fault diagnosis system． Keywords：active power filter(APF)，feature extraction，fault diagnosis，wavelet packet，neural network 0 引 言 需要对APF进行状态监测与诊断。APF中的功率管 随着现代高科技的迅猛发展，大量的变频变流