基于BP神经网络的光伏并网逆变器控制方法研究  

Research on the Control Method of Photovoltaic Grid-connected Inverter Based on BP Neural Network

作　　者：范其丽[1] 王璞 冯越 FAN Qili;WANG Pu;FENG Yue(Department of Electrical and Electronic Engineering,Zhengzhou Technical College,Zhengzhou 450121,Henan,China;State Grid Zhumadian Power Supply Company,Zhumadian 463000,Henan,China)

机构地区：[1]郑州职业技术学院电气电子工程系,河南郑州450121 [2]国网驻马店供电公司,河南驻马店463000

出　　处：《电气传动》2020年第4期59-62,共4页Electric Drive

摘　　要：以光伏并网逆变器为研究对象,为解决非线性可变负载并网时动态响应慢、电压波动大等问题,基于BP神经网络提出了一种逆变器控制方法。详细论述了光伏并网逆变器主电路结构以及数学模型。针对并网电流内环控制,设计了一种BP神经网络的控制器。在保证输出误差最小的前提下,采用梯度下降法寻找PID参数最优值,实现PID参数的实时调整。通过调整网络权值和学习率消除负载变化造成的不利影响,加快系统响应。仿真结果表明:在BP-PID控制策略下,并网电流跟踪速度更快、效果更好,可基本确保电流误差稳定在零附近,较好地完成了并网电流跟踪,验证了该控制策略的可靠性。Taking photovoltaic grid-connected inverter as the research object,in order to solve the problems of slow dynamic response and large voltage fluctuation,an inverter control method based on BP neural network was proposed.The main circuit structure and mathematic model of photovoltaic grid-connected inverter were discussed in detail.A BP neural network controller was designed for the control of the inner loop of grid-connected current.On the premise of ensuring the minimum output error,the gradient descent method was adopted to find the optimal value of PID parameters and the real-time adjustment of PID parameters was also realized.By adjusting the network weight and learning rate to eliminate the negative impact of load change and speed up the system response.The simulation results show that under the BP-PID control strategy,the grid-connected current tracking speed is faster and the effect is better,which can basically ensure that the current error is stable near zero and the grid-connected current tracking is completed well,which verifies the reliability of the control strategy.

关 键 词：逆变器 光伏并网 神经网络 PID控制 

分 类 号：TM743[电气工程—电力系统及自动化]