Fault diagnosis of analog circuits based on wavelet neural network

A fault diagnosis method for analog circuits was proposed. Wavelet neural network was used to detect the faults and classify default patterns in circuits under test. To construct this new network, wavelet functions were embedded in a neural network as neuron stimulating functions, which combined the multi-scale analysis merit of wavelet transform with nonlinear mapping capability of neural network. Wavelet multi-resolution decomposition was adopted to extract fault features after sampling the impulse response of the analog circuit. The features were applied to train the wavelet neural network for further fault patterns recognition. Diagnosis principle and procedure were elucidated. Simulation results on band pass filter circuit show that the presented approach has good capability of fault identification and diagnosis.