• CSCD核心库收录期刊
  • 中文核心期刊
  • 中国科技核心期刊

电力建设 ›› 2019, Vol. 40 ›› Issue (12): 113-119.doi: 10.3969/j.issn.1000-7229.2019.12.014

• 智能电网 • 上一篇    下一篇

基于小波变换与BP神经网络的光伏电站孤岛检测研究

江文超1,张兴1,谢东2,李明1   

  1. 1.合肥工业大学电气与自动化工程学院,合肥市 230009;2.铜陵学院电气工程学院,安徽省铜陵市 244000
  • 出版日期:2019-12-01
  • 作者简介:江文超(1995),男,硕士研究生,主要研究方向为新能源发电技术; 张兴(1963),男,博士,教授,通信作者,主要研究方向为新能源利用与分布式发电技术; 李明(1993),男,博士研究生,主要研究方向为新能源利用与分布式发电技术; 谢东(1980),男,博士,副教授,主要研究方向为新能源及分步式发电技术。
  • 基金资助:
     国家重点研发计划项目(2016YFB0900300)

Research on Photovoltaic Power Station Islanding Detection Based on Wavelet Transform and BP Neural Network

JIANG Wenchao1, ZHANG Xing1 , XIE Dong2, LI Ming1   

  1. 1.School of Electric and Automatic Engineering, Hefei University of Technology, Hefei 230009, China;2. School of Electric Engineering,Tongling University, Tongling 244000, Anhui Province, China
  • Online:2019-12-01
  • Supported by:
    This work is supported by National Key Research and Development Program of China(No.2016YFB0900300).

摘要: 随着光伏电站规模不断增大,并网光伏系统对原有供电网络的影响越来越大,孤岛检测成为光伏电站必须深入研究的问题。针对现有孤岛检测方法的不足,提出了一种基于小波变换与BP神经网络的新型被动式孤岛检测法。该法通过小波变换获得有关信号孤岛发生前后的特征信息,再由BP神经网络根据这些特征信息实施孤岛检测和孤岛保护行为。仿真研究的结果表明,所述新型被动式孤岛检测方法检测速度快,检测盲区小,在多个负载品质因数、谐波等扰动情况下,不会出现孤岛检测的误判行为。

关键词: 孤岛检测, 小波变换, 神经网络, 特征向量

Abstract: With the increasing scale of photovoltaic power station and the impact of grid-connected photovoltaic systems on the original power supply network, islanding detection has become a problem that PV power station must study in depth. Aiming at the deficiencies of the existing islanding detection methods, a novel passive islanding detection method based on wavelet transform and BP neural network is proposed. The method obtains the characteristic information of the signal before and after islanding through wavelet transform, and then the BP neural network implements islanding detection and islanding protection behavior according to this characteristic information. The simulation results show that the new passive islanding detection method described in this paper has high detection speed and small no-detection zone. In the case of multiple load quality factors and harmonics, no misjudgment of islanding detection will occur.

Key words: islanding detection, wavelet transform, neural network, eigenvector

中图分类号: