A Novel Fault Diagnosis Based Improved WOA-BP for Analog Circuit

Aiming at the problem that the fault diagnosis signals in the existing analog circuits cannot be effectively classified, a new scheme (BL-WOA-BP) based on the improved whale algorithm(WOA) to optimize the parameters of the back-propagation(BP) neural network is proposed in this paper. The scheme selects the wavelet packet analysis(WPA) with higher precision than the traditional wavelet analysis to extract the fault features of the analog circuit; Then, the feedback mechanism is used to improve the global search ability of the WOA, and the transformation of linear and nonlinear factors is used to balance the local and global search ability, so as to better optimize the parameters of the BP neural network. Finally, three analog circuits of different scales are trained and tested by establishing BL-WOA-BP model to realize circuit soft fault diagnosis. After verifying the scheme through simulation experiments, it is shown that the improved WOA optimized BP neural network has higher diagnostic accuracy compared with other methods compared in this paper, and the fault recognition rate reaches 100% at most. Through the neural network based on embedded system, the real-time fault diagnosis of analog circuit is carried out to diagnose the circuit fault timely and effectively.