基于模糊SVM和虚拟仪器的模拟电路故障诊断研究  

Fault Diagnosis Method of Analog Circuits Based on Virtual Instrument and Fuzzy SVM

作　　者：邓森[1] 杨军锋[1] 郭明威[1] 郭创[1] 

机构地区：[1]空军工程大学工程学院,陕西西安710038

出　　处：《计算机测量与控制》2011年第4期762-765,805,共5页Computer Measurement &Control

基　　金：航空科学基金(20080896009)

摘　　要：基于模糊支持向量机理论构建模拟电路故障诊断网络,采用虚拟仪器技术开发故障诊断平台;通过对电路仿真软件与实际测量得到的数据进行分析,选取一种自适应小波变换特征提取方法对电路进行故障特征提取,提取电路输出响应的6个低频系数构成故障特征向量并作为FSVM诊断网络的学习样本,诊断网络采用C-SVM算法,规则化参数取为200;在LabVIEW软件中调用以MATLAB.M文件编写的特征提取与故障诊断算法,将模拟电路的故障定位到元件级;最后,将网络的诊断结果与BP神经网络诊断方法做了对比,证明基于虚拟仪器的模糊SVM模拟电路诊断方法在故障诊断速度与准确性方面都具有明显优势,平均故障识别率达到90%以上。Based on virtual instrument and fuzzy SVM,a fault-diagnosis system was set up for analog circuits.According to analysis of data from circuit emulation software and practical measurement,the fault characters of circuit output were extracted by using the way of self adaptive Wavelet.The fault characteristic vectors were made of 6 low frequency coefficients and the learning samples for fuzzy SVM were also got by these.In fault-diagnosis system,C-SVM algorithm was used and C parameter was 200.Using the MATLAB M-file about characteristic extraction and fault diagnosis algorithm in LabVIEW,the fault-diagnosis of analog circuits can be detected in components level.At last,the result of fault diagnosis of BP neural-network was compared with that of fuzzy SVM.The result shows that fuzzy SVM method has a transparent superior in fault diagnosis with high speed and accuracy,the average rate of fault recognition can reach 90%.

关 键 词：模糊支持向量机 虚拟仪器 模拟电路 故障诊断 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]