Fault Diagnosis of Analog Circuits Based on Wavelet Packet Energy Entropy and DBN

Abstract

Abstract To solve the problem of low classification and recognition rate caused by small number of fault samples and inaccurate feature extraction in analog circuit fault diagnosis, a fault diagnosis algorithm for analog circuit system based on the combination of wavelet packet energy entropy and Deep Belief Network (DBN) is proposed. Firstly, the original output voltage signal of the circuit is decomposed by a multi-layer wavelet packet, then the feature vector is constructed in the form of energy entropy, and then the principal component analysis (PCA) is used for feature selection. The reduced dimension feature vector is taken as the input vector of DBN, and the fault diagnosis is completed after training and learning of the DBN network model. The experimental results show that compared with other algorithms, the proposed method can be more accurate and effective in diagnosing fault types in analog circuits, especially for Sallen-Key band-pass filter circuits, the fault recognition rate reaches 100%.