核函数加权FCM聚类算法下的电力电子电路故障诊断

Fault Diagnosis of Power Electronic Circuit Based on Weighted Fuzzy-C Mean Kernel Clustering Algorithm

电力电子电路故障检测和诊断,是保证电力电子装置安全、可靠运行的一个重要技术手段.针对传统电力电子电路故障诊断方法存在的缺陷,提出了基于复相关系数的加权核函数的FCM聚类方法,并将用于对Cuk电路的故障分析,使之不但能够达到比较好的分类效果,同时也适用于样本分类不均衡的聚类问题,具有较高的诊断率和较低的误诊断率,且故障诊断的时间比较短.为电力电子电路故障的初期诊断和故障数据处理提供了一个有效的方法和途径.

The fault detection and diagnosis of power electronic circuit can ensure circuit running reliably and safely.The fault diagnosis for power electronic circuit method is limited,so a type of fault diagnosis system for Cuk circuit based on weighted fuzzy-c mean kernel clustering algorithm is designed.It can not only enhance classification performances,but also be used in the sample clustering problem of unequal classification,with better diagnosis ratio,lower misdiagnosis ratio and shorter treatment time.It provides an effective fault diagnosis method for the incipient diagnosis and data processing of power electronic circuit.