基于经验模态分解与Elman神经网络的永富直流换相失败故障诊断方法  

Yongfu DC commutation failure fault diagnosis based on the EMD-Elman neural network

作　　者：杨鸿雁 邢超 陈仕龙[1] 严增伟 刘浩 Yang Hongyan;Xing Chao;Chen Shilong;Yan Zengwei;Liu Hao(College of Electrical Engineering,Kunming University of Science and Technology,Kunming 650500,China;Yunnan Electric Power Research Insititute,Yunnan Power Grid Co.,Ltd.,Kunming 650011,China;Kunming Power Supply Bureau Yunnan Power Grid Co.,Ltd.,Kunming 650011,China)

机构地区：[1]昆明理工大学电力工程学院,昆明650500 [2]云南电网有限责任公司电力科学研究院,昆明650011 [3]云南电网有限责任公司昆明供电局,昆明650011

出　　处：《电子测量技术》2020年第1期169-175,共7页Electronic Measurement Technology

基　　金：国家自然科学基金项目(51267008)资助。

摘　　要：直流输电系统因自身的优越性能已广泛应用于电力系统。换相失败是直流输电系统中一种常见故障,会导致电流电压发生突变,给直流系统的安全稳定运行带来严重威胁。利用PSCAD/EMTDC平台搭建了永富直流输电系统模型,提出了基于经验模态分解和Elman神经网络相结合(EMD-Elman)的换相失败故障诊断方法。通过大量仿真实验发现使用Elman+样本熵可以有效的进行故障诊断,仅在50组样本的情况下对换相失败故障识别的精确度就可以达到95%以上,证明了该换相失败识别方法的有效性。HVDC transmission systems have been widely used in power systems due to their superior performance. Commutation failure is a common fault in the DC transmission system, which will cause sudden changes in current and voltage, which poses a serious threat to the safe and stable operation of the DC system. In this paper, the model of Yongfu HVDC transmission system is built by PSCAD/EMTDC, and the fault diagnosis method of commutation failure based on EMD-Elman neural network is proposed. The commutation failures in different cases are analyzed, and the neural network is combined with the approximate entropy and sample entropy for identification. Through a large number of simulation experiments, it is found that the use of Elman+ sample entropy can effectively diagnose faults, accuracy of commutation failure fault identification can reach over 95% in only 50 sets of samples, prove the effectiveness of the commutation failure identification method.

关 键 词：永富直流 经验模态分解 ELMAN神经网络 换相失败 故障诊断 

分 类 号：TP771[自动化与计算机技术—检测技术与自动化装置]