为了提高相控开关动作时间预测精度,抑制电容器投切产生的过电压和涌流,文章建立了以控制电压和环境温度为输入的前馈网络预测模型;为了提高模型预测精度,提出基于遗传算法(genetic algorithm,GA)和粒子群算法优化神经网络的补偿方法,并对算法优化前、后网络预测性能进行比较。研究结果表明,经过遗传算法和粒子群优化后的前向神经网络模型比没有优化的有更好的预测精度。
To improve the action time forecasting accuracy in phase-controlled swltclllng, ano suppres~ the overvoltage and inrush current generated by capacitor switching, an operating time forecasting model is developed. This model is a BP neural network with the control voltage and temperature as in- put variables. To improve the action time forecasting accuracy o~ the model, the optimization method of neural network with genetic algorithm(GA) or particle swarm optimization(PSO) is proposed. The performance of the neural network model with GA or PSO is compared with that without optimiza- tion. The research results show that the BP neural network model optimized with GA or PSO posses- ses better forecasting accuracy than that without optimization.