Soft Fault WNN Diagnosis Method for Analog Circuit Based on Improved AFSA

Abstract

To improve the soft fault diagnosis ability of analog circuits, a wavelet network (WNN) soft fault diagnosis method based on improved Artificial Fish Swarm Algorithm (AFSA) is proposed. In the method, chaos initialization strategy and mutation factor are introduced to improve the shortcomings of standard AFSA, so as to improve the global search ability of WNN parameters and overcome local convergence. According to the soft fault diagnosis flow of the designed analog circuit, the WNN model trained by improved AFSA is built. Simulation shows that the method is effective and feasible for soft fault diagnosis of analog circuits.