基于模糊神经网络的特高压输电系统换相故障诊断方法分析  

Analysis of Commutation Fault Diagnosis Method of UHV Transmission System Based on Fuzzy Neural Network

作　　者：孙飞 孔伟伟 付成成 刘清 SUN Fei;KONG Weiwei;FU Chengcheng;LIU Qing(State Grid Communication Co.,Ltd.Anhui Jiyuan Software Co.,Ltd.,Anhui 230022,China)

机构地区：[1]国网信通股份公司安徽继远软件有限公司,安徽230022

出　　处：《电子技术（上海）》2023年第10期46-49,共4页Electronic Technology

摘　　要：阐述一种基于模糊神经网络的特高压输电系统换相故障诊断方法,该方法利用霍尔电流传感器和示波器组成的采集装置,采集特高压输电系统的直流电流信号,并实施去噪处理。利用集合经验模态分解(Ensemble Empirical Mode Decomposition, EEMD)分解直流电流信号,计算固有模态函数(Intrinsic Mode Function, IMF)分量近似熵和样本熵,并实施归一化处理,组成换相故障特征向量。输入提取的换相故障特征向量,通过模糊神经网络实现故障诊断,确定故障类型。结果表明,在所研究诊断方法应用中,交并比大于对比方法的故障诊断结果,说明所研究诊断方法准确性更高。This paper describes a fuzzy neural network-based commutation fault diagnosis method for UHV transmission systems,which uses a collection device composed of a Hall current sensor and an osilloscope to collect the DC current signal of the UHV power transmission system and implement denoising processing.Ensemble Empirical Mode Decomposition(EEMD)is used to decompose the DC current signal.The approximate entropy and sample entropy of the Intrinsic Mode Function(IMF)component are calculated,and normalization is performed to form the commutation fault feature vector.Input the extracted commutation fault feature vector.Realize fault diagnosis through fuzzy neural network and determine the fault type.The results show that under the application of the researched diagnosis method,the combined ratio is greater than the fault diagnosis result of the contrast method,indicating that the researched diagnosis method is more accurate.

关 键 词：模糊神经网络 特高压输电系统 直流电流信号 去噪 故障特征 诊断方法 

分 类 号：TP126.45[自动化与计算机技术—控制理论与控制工程] TP183[自动化与计算机技术—控制科学与工程]