Analog circuit fault diagnosis combing wavelet packet with higher order statistics

An approach to fault diagnosis for analog circuit is described in this paper, which uses wavelet packets transform and higher order statistics (HOS) method to extract fault features and SVM classifiers to classify faults. Firstly, output voltage signals are obtained from the test nodes and the fault feature vectors are extracted using wavelet packets transform and HOS method. Then, the fault feature vectors are feed to SVM for classification. Simulation results of diagnosing a four op-amp high pass filter show us the proposed method has the higher classification accuracy and have confirmed the validity of the proposed method.