A Current Zero-Crossing Distortion Compensation Method Based on BP Neural Network In Fast-Scanning Power Supply

fast scanning power supply is utilized in many accelerator fields. Current waveform distortion caused by various reasons often occurs at the zero-crossing point, which reduce accuracy of output current. In order to optimize the distortion of current waveform, a waveform compensation method based on BP neural network of Bayesian regularization is introduced. Firstly, the reasons for the current zero-crossing distortion of fast scanning power supply are analyzed theoretically. Then a principle of optimization method is proposed and implemented. Finally, verification of compensation method is carried out on a sine-wave scanning power supply. The result shows that the compensation method can effectively suppress the current distortion. And it is also conducive to achieving the automation of power supplies.