Wavelet neural networks based fault diagnosis of analog circuit

The fault diagnosis of analog circuits has made many ways, such as the fault dictionary method, device parameter identification method and so on. However, due to the diversity of fault types, component tolerances, nonlinear, and the measured response signal usually contains non-stationary and time-varying signal component in the analog circuit failure, the fault diagnosis becomes difficult. Fault diagnosis method based on wavelet neural network can solve these problems. Multisim software is used to establish analog circuits, wavelet neural network is trained by additional momentum term and adaptive learning rate BP algorithm, the characteristics of analog circuit fault is classified and identified, simulation results demonstrate the effectiveness of the method.