克隆选择优化的SVM模拟电路故障诊断方法

对基于克隆选择算法的支持向量机(SVM)参数优化、及其在模拟电路故障诊断中的应用进行了深入研究,故障诊断实现步骤为:首先对电路的各种故障模式进行蒙特卡洛仿真分析,利用小波分解提取输出信号的各频段能量,进行归一化处理后得到故障特征样本;然后应用克隆选择算法进行SVM参数优化,并将选定的参数用于SVM的训练;最后采用训练好的SVM对故障样本进行分类,从而实现故障判定。论文以CTSV滤波电路和螺距反馈电路为诊断实例进行了实验验证,结果表明对容差模拟电路的故障定位具有较高的准确率。

Analog circuit fault diagnosis based on SVM optimized by CSA

A method for fault diagnosis of analog circuit based on Support Vector Machines（SVM） optimized by Clonal Selection Algorithm（CSA） is presented in this paper.Circuit fault samples are extracted from the result of Monte-Carlo simulation on each fault class by using wavelet analysis.And the training samples and SVM parameters op-timized based on CSA are used for SVM training.Then the fault samples are diagnosed by use of the trained SVM.In the end of this paper,experiments on a CTSV filter circuit and pitch feedback circuit are carried out.And the result indicates that the proposed method has the capability to diagnose faults in tolerance circuits and achieves satisfactory accuracy.