Online Excitation Current Pre-Charge Time Adjustment of Output Current Waveform for Pulsed Alternator Based on Backpropagation Neural Network

Abstract

In this article, a method based on backpropagation (BP) neural network is proposed to adjust the output current waveform by adjusting the excitation current pre-charge time online. The discharge controller adjusts the output current waveform of the pulsed alternator by automatically modifying the excitation current pre-charge time according to the different current waveform requirements of the load. First, the characteristics of the pulsed alternator in the pulse separate excitation mode are analyzed, and the intrinsic characteristics of the excitation circuit are highlighted. Subsequently, based on the strong nonlinear relationship between different excitation current pre-charge time and their corresponding output pulse waveforms, a BP neural network model is constructed, mapping the output pulse waveform indices to the excitation current pre-charge time. Second, a 3-D field-circuit coupling finite element model of the pulsed alternator is established, and suitable samples for neural network training are collected using this model, facilitating the training of the neural network. Finally, the correctness and effectiveness of the proposed method are verified through experimental research conducted on a prototype platform of the pulsed alternator.