基于双空间的融合特权信息支持向量机模拟电路故障诊断

针对模拟电路故障诊断识别率较低的问题,提出了基于双空间特征提取的融合特权信息支持向量机的模拟电路故障诊断新方法。首先对采集的信号进行主成分分析(principal component analysis,PCA)——特征提取;并用融合特权信息支持向量机LUPI-SVM(SVM of learning using privileged information,LUPI-SVM)分类器和SVM-GA分类器进行预分类;对分类结果不同的样本进行独立成分分析(independent component analysis,ICA)—特征提取,并用LUPI_SVM进行分类识别,Sallen-Key滤波电路故障诊断仿真实验结果表明该方法有效提高了分类的性能,为模拟电路故障诊断提供了新的途径。

A novel analog circuit fault diagnosis method based on dual-space feature extraction algorithm and SVM of learning using privileged information

a novel fault diagnosis method based on dual-space feature extraction algorithm and LUPI-SVM is proposed in this paper.,Aiming at solving the problem of correctly identifying fault classes in analog circuit fault diagnosis.Firstly,the fault feature vectors are extracted by PCA feature extraction method.Then,pre-classfy them by the LUPI_SVM classfier and the SVM-GA classfier,and then the different vectors of the pre-claasfying results are extracted by ICA.Finally,inputing them into the trained LUPI-SVM model to identify the different fault cases.The simulation results for analog and mixed-signal test benchmark Sallen-Key filter circuits demonstrate that the proposed method improved classification ability correctly,which develops a new direction for the fault diagnosis of analog circuit.