基于小波分解和模糊聚类的模拟电路软故障诊断

Analog Circuit Soft Fault Diagnosis Based on Wavelet Analysis and Fuzzy Clustering

为了解决模拟电路故障诊断中出现的故障特征冗余的问题,提出一种采用小波分解和模糊聚类提取故障特征的模拟电路故障诊断方法.对被测电路的输出进行小波分解,将得到的重构小波系数作为故障特征而构成测试矩阵,再利用基于马氏距离的模糊聚类的方法对测试矩阵进行划分,降低测试矩阵的维数,最后用BP神经网络进行故障诊断.仿真实验结果表明,所提出的方法能有效提取出故障特征,并且具有较高的模拟电路故障诊断准确率.

A method of analog circuit fault diagnosis based on extracting fault features by using wavelet decomposition and fuzzy clustering is presented in this paper.This approach performs wavelet decomposition on the acquired output signals of the test circuit and takes the reconstructing wavelet coefficients as the fault feature values,then using the method of fuzzy clustering based on markov distance to division the test matrix to reduce the dimensions of the test matrix,finally the obtained test matrix are inputted into the neural network for fault diagnosis.The simulation result show 是 that the proposed method can effectively extract the fault feature and obtain well accuracy of analog circuit fault diagnosis.