针对级联式变频器内部功率管开路故障诊断中,逆变侧功率管开路故障隐蔽性较强、诊断较难的问题,提出了一种基于小波包特征熵的故障信号提取方法.为了提高级联式变频器功率管开路故障的诊断精度,采用马氏距离分类法进行故障诊断.首先,采集某型号级联式变频器在不同工况下的输入侧电流信号;其次,对采集的电流信号作小波包变换,并提取其特征熵向量作为样本数据集;最后,利用马氏距离分类法进行故障诊断.试验结果表明:采用小波包特征熵提取算法,可以有效地提取级联式变频器功率管发生开路故障时的电流信号特征;同时,采用马氏距离分类法,能够较好地对特征熵向量进行分类和识别.2种算法的结合,可以有效诊断级联式变频器功率管开路故障,也为变频器功率管开路故障的诊断提供了新方法.%The concealment of the open circuit fault of power tube in inverter side of the cascade frequency converter is strong, and the fault diagnosis is difficult. Aiming at this problem, the extraction method of the fault signals by using wavelet packet characteristic entropy is proposed. In order to improve the accuracy of diagnosis,the Mahalanobis distance classification is adopted for fault diagnosis. Firstly,the current signals of the input side of cascade converter in different working conditions are collected;then,the current signals collected are transformed by the wavelet packet,to extract the characteristic entropy vectors as sample data set;finally,the method of Mahalanobis distance classification is used to diagnose the fault. The experiment results show that the method of signal extraction based on the wavelet packet characteristic entropy can effectively extract the characteristics of the fault current signal,and the method of Mahalanobis distance can well classify and identify the characteristic entropy vectors. The combination of the two algorithms can effectively diagnosis the open circuit fault of power tube in cascade inverter. It also provides a new method for the diagnosis of the power tube open circuit faults in converter.
展开▼
