基于BP神经网络的并联型有源电力滤波器直流侧电压PID控制设计

摘要

Based on the traditional PI control method for active power filter DC-side voltage, targeting the deficiency of the PI control method,this paper proposes the PID control method based on BP neural network. This method makes full use of the ability of approaching random nonlinear function,and has the trait of rapid dynamic response,small overshoot and small static state error. Furthermore, it is easy to carry out and can satisfy the demand for load versatility to DC voltage of shunt active power filter. The structure of BP neural network is designed and the weighted coefficient calculation formula of hidden layer and output layer is derived, with the flow chart for the learning algorithm of the BP neural network given. Simulation studies on the control effect are made through Matlab.%在传统的基于PI控制的有源电力滤波器直流侧电压控制方法的基础上,针对PI控制方法的不足,提出了基于BP神经网络的并联型有源电力滤波器直流侧电压的PID控制.该方法充分发挥了神经网络具有逼近任意非线性函数的能力,具有动态响应快、超调小、静态误差小的特点,并且实现简单,满足多种变化负载情况下的有源电力滤波器直流侧电压控制要求.对BP神经网络结构进行了设计并推导出隐含层和输出层加权系数计算公式,同时给出了BP神经网络学习算法流程图.通过Matlab对其控制效果进行了仿真研究.