SCA-SVM Fault Diagnosis of Analog Circuits Based on Transfer Learning

To solve the problem that there are Lack of sufficient fault samples in circuit fault diagnosis methods meanwhile generations for obtaining failure samples are too expensive is proposed, this paper proposes sine and cosine algorithm and support vector machine (SCA-SVM), a challenging research method of circuit fault diagnosis, that is, modeling is carried out under the premise of using a small number of target fault samples. The algorithm uses the idea of transfer learning to diagnose the fault of analog circuit. First, the feature is extracted from the data by wavelet packet, then the dimension is reduced by principal component analysis (PCA), then the fault sets are classified by SCA-SVM, and finally the accuracy is determined by Sallen-Key band pass filter circuit and CSTV filter circuit. Overall, compared with back-propagation algorithm (BP) genetic algorithm (GA) learning vector quantization algorithm (LVQ) and PCA, SCA performance in few-shot learning (FSL) is better than other algorithms.