神经网络准PR光伏并网逆变器控制技术  

Quasi PR Photovoltaic Grid-connected Inverter Control Method Based on BP Neural Network

作　　者：范宝奇 罗晓曙[1] 廖志贤[1] 姚鑫[1] 

机构地区：[1]广西师范大学电子工程学院,桂林541004

出　　处：《电力系统及其自动化学报》2016年第3期30-34,共5页Proceedings of the CSU-EPSA

摘　　要：针对单相并网逆变系统高度非线性的特性,为解决传统逆变器控制系统自适应能力差的问题,在分析了比例谐振PR(proportional resonant)控制与准PR控制策略的优缺点的基础上,将神经网络算法和准PR算法结合,提出一种基于神经网络参数自整定的准PR控制方法。解决了准PR控制数字化精度不够和参数整定困难的问题。利用Matlab/Simulink平台对神经网络准PR控制进行仿真,仿真结果表明:与准PR控制相比,基于BP神经网络准PR控制的电流跟踪总谐波畸变率降低,动态响应性能更快,系统自适应程度更高,有较好的应用价值。According to the highly nonlinear characteristics of single-phase grid connected inverter system, in order to solve the poor adaptive capacity of the traditional inverter, based on the analysis of the advantages and disadvantages of quasi-PR control and PR control strategy, the neural network algorithm and quasi-PR algorithm are combined, a self- tuning parameter neural network quasi-PR control method is put forward. The problem of the imprecise quasi PR digital control and parameters tuning difficult is be solved. Using the Matlab/simulink platform train, the neural network PR control is simulated. The simulation results show that compared with quasi PR control, the total harmonic distortion of tracking current which produced by neural network quasi PR control method decreased, and the dynamic response per- formance is better, and the adaptive degree of the system became higher. So this method has great application value.

关 键 词：并网逆变器 误差反传神经网络 电流控制 准比例谐振控制 

分 类 号：TM615[电气工程—电力系统及自动化]