新型复合神经网络控制的并联有源电力滤波器

Novel Shunt Active Power Filter Based on Composite Neural Network

为向用户提供高质量的电能,同时消除负载对供电系统的不利影响,提出了一种新的基于复合神经网络的并联有源电力滤波器(APF),并将其应用于配电系统的谐波治理中。采用复合神经网络组成APF提取控制回路,它分为补偿信号提取回路和逆变器控制回路两部分。前者确定电流中的谐波成分,后者通过对PWM逆变器的输入信号的调整使逆变器提供更精确的补偿。仿真结果表明,复合神经网络控制的有源电力滤波器能够显著降低供电系统中的畸变,且反应迅速,补偿精度高,效果稳定。

This paper proposes a novel composite neural network controlled active power filter （APF） for harmonic compensation. This composite network includes a compensation extraction net and an inverter controller net. The harmonics of load current, which are used to control the inverter, are computed by the extraction net. The output of the inverter is adjusted by the inverter controller net. Simulation results show that the composite neural network controlled active power filter can eliminate power system harmonics quickly and effectively.