基于小波变换和神经网络的光伏发电孤岛效应检测方法  

A Method Detecting the Island Effect of the Photovoltaic Power Generation Based on the Wavelet Transform and Neural Network

作　　者：郗忠梅[1,2] 张世恒 朱瑞婷[4] 李胜杰 刘双喜[1,2] 彭宏[6] 

机构地区：[1]山东农业大学机械与电子工程学院,山东泰安271018 [2]山东农业大学山东省园艺机械与装备重点实验室,山东泰安271018 [3]国网西宁供电公司,青海西宁810008 [4]上海电力学院电气工程学院,上海200090 [5]国网山东茌平县供电公司,山东茌平252100 [6]国网天津市电力公司城东供电分公司,天津300250

出　　处：《山东农业大学学报（自然科学版）》2016年第2期254-258,共5页Journal of Shandong Agricultural University：Natural Science Edition

摘　　要：针对分布式发电系统的孤岛现象检测,本文提出了基于小波分析和神经网络的孤岛效应检测方法。该方法对公共耦合点的电压采样,并经小波分解后得到电压变化的细节分量,将其输入神经网络进行学习、训练。测试结果表明,小波神经网络的检测方法正确率可达98%以上。当孤岛效应发生时,该方法能够快速准确地检测出来,既解决了对电网造成谐波污染的问题,又减小了不可检测域的存在。Based on the wavelet analysis and neural network, a method detecting the island effect is presented in this paper to detect the island phenomenon for distributed power generation system. The method first involves sampling the common coupling point voltage and decomposing the wavelet, then the detailed weight of the voltage change could be obtained and input into the neural network tbr learning and training. The results proves that the accuracy of detection method of wavelet neural network could be more than 98%. The occurrence of island effect can be detected duly and precisely, thus tackling the problem of harmonic pollution to power network and removin,a the undetectable dornain.

关 键 词：光伏逆变 孤岛效应 小波变换 神经网络 

分 类 号：TM615[电气工程—电力系统及自动化]