基于WPT-VMD-BP的孤岛检测方法
An islanding detection method based on WPT-VMD-BP
投稿时间:2022-03-01  修订日期:2022-03-01
DOI:
中文关键词: 被动孤岛检测法  小波包去噪  变分模态分解  BP神经网络  分布式电站
英文关键词: passive islanding detection  wavelet packet transform  variational mode decomposition  BP neural network  distributed generator
基金项目:国家自然科学基金项目(面上项目,重点项目,重大项目)
作者单位邮编
王增雯 湖北工业大学 
黄文聪 湖北工业大学 430068
常雨芳 湖北工业大学 
摘要点击次数: 95
全文下载次数: 0
中文摘要:
      针对现有被动孤岛检测方法检测速度慢、检测准确率不高的缺陷,提出一种小波包(WPT)去噪、变分模态分解(VMD)和BP神经网络相结合的孤岛检测方法。首先对分布式光伏电站不同运行工况公共耦合点(PCC)处的电压信号进行去噪预处理;然后对该信号进行变分模态分解,得到具有不同中心频率的模态分量,再将其合成分类所需的电压特征向量;最后利用BP神经网络对不同电压特征向量进行学习分类,进而识别孤岛状态。通过PSCAD/MATLAB联合仿真,对所提方法检测效果进行了验证,并探究了不同干扰工况下孤岛检测方法的抗干扰性能。
英文摘要:
      An intelligent islanding detection method based on the combination of wavelet packet transform denoising(WPT), variational mode decomposition(VMD) and Extreme Learning Machine(ELM) is proposed aiming at solving the problem of slow speed and low accuracy during passive islanding detection. Firstly, PCC voltage signals at PV station in different operation condition are denoised by WPT. Secondly, the voltage signal are decomposed by variational modal decomposition method, then the voltage vectors containing a large number of islanding features are obtained by introducing the concept of Shannon entropy. Finally, the extreme learning machine is used to classify and recognize islanding conditions. The results of PSCAD/MATLAB simulation show the performance of proposed method and the proposed method has small non-detection areas, fast detection speed, high reliability even under disturbing operation conditions.
View Fulltext   查看/发表评论  下载PDF阅读器
关闭