基于压缩感知和指纹图谱的变压器局部放电空间定位方法  

Spatial Location Method of Partial Discharge in Power Transformer Using Compressed Sensing and Fingerprinting

作　　者：冯晓泽 寇鹏[1] 梁得亮[1] 崔新奇 王浩洋 FENG Xiaoze;KOU Peng;LIANG Deiang;CUI Xinqi;WANG Haoyang(School of Electrical Engineering of Xi’an Jiaotong University,Xi’an 710049,China;The State Grid Shanghai Maintenance Company,Shanghai 200063,China)

机构地区：[1]西安交通大学电气工程学院,西安710049 [2]国网上海市电力公司检修公司,上海200063

出　　处：《高压电器》2023年第12期160-167,共8页High Voltage Apparatus

基　　金：国家电网有限公司科技项目资助(5500-201999543A-0-0-00)。

摘　　要：电力变压器作为电网的关键承载点,若发生故障将严重影响电网的安全运行。绝缘故障在变压器故障中占比最高,而局部放电(partialdischarge,PD)又是导致绝缘故障的主要原因,因此对PD进行有效定位有重要实际意义。传统的变压器超高频PD定位方法主要基于到达时间差的分析,但其噪声敏感性以及高同步采样要求所带来的硬件开销限制了应用。对此,文中提出一种基于压缩感知和接收信号强度(RSSI)指纹图谱的变压器局部放电空间定位方法。该方法主要包括两个步骤:基于近邻传播聚类和神经网络的初定位,以及基于压缩感知的精确定位。仿真结果表明平均定位误差为0.017m,验证了该方法具有较高的精确性与实用价值。Power transformer is the key load carrying point of the power gird and any faults in it can will affect seri-ously the safe operation of power grid.Insulation faults account for the highest proportion of transformer faults,and partial discharge(PD)is the main cause leading to insulation fault.Therefore,it is of great practical significance to effectively locate the PD.As for power transformers,traditional ultra high frequency(UHF)PD localization method is mainly based on the analysis of arrival time difference.However,its noise sensitivity and hardware overhead due to high synchronous sampling requirements limit its application.To address this,a PD spatial localization method us-ing compression sensing and the received signal strength(RSSI)fingerprinting is proposed in this paper.This method mainly includes two steps.One is the initial localization based on affinity propagation clustering and neural network,and the other is the precise localization based on compression sensing.Simulation results show that the average loca-tion error is 0.017 m and the method is verified to have higher accuracy and practical value.

关 键 词：变压器局部放电 压缩感知 近邻传播算法 神经网络 指纹图 空间定位 

分 类 号：TM855[电气工程—高电压与绝缘技术] TM41