基于有载分接开关的变压器绕组故障自动化诊断方法研究  

Study on the automatic diagnosis method of transformer winding faultbased on on-load tap switch

作　　者：朱宜东 吕志勇 黄文彬 ZHU Yidong;LYU Zhiyong;HUANG Wenbin(Zaozhuang Power Supply Company State Grid Shandong Electric Power Company,Zaozhuang 277100,China)

机构地区：[1]国网山东省电力公司枣庄供电公司,山东枣庄277100

出　　处：《无线互联科技》2024年第21期33-35,共3页Wireless Internet Science and Technology

摘　　要：变压器绕组故障信号状态的多样性对故障诊断效果具有显著影响,因此,文章提出基于有载分接开关的变压器绕组故障自动化诊断方法。该方法将单维时间序列下的信号映射到高维空间,通过利用信号能量熵函数,确定相点在映射空间中的累积分布特征。引入长短期记忆网络技术,分析有载分接开关信号累积分布的延迟特性,基于提取的特征自动进行故障模式的识别。测试结果表明:该研究方法对于各类变压器绕组故障的诊断用时始终在2.5 ms以下,诊断性能可靠性较高。Since the diversity of transformer winding fault signal status influences on the fault diagnosis effect,the automatic diagnosis method of transformer winding fault diagnosis based on on-load tap switch is proposed.This method maps the signals under the single-dimensional time series to a high-dimensional space,and determines the cumulative distribution characteristics of the phase points in the mapping space by using the signal energy entropy function.Long-and short-term memory network technology is introduced to analyze the delay characteristics of the cumulative distribution of the on-load tap switch signals,and automatically identify the fault mode based on the extracted features.The test results show that the various transformer winding fault diagnosis time is always under 2.5 ms,and the diagnostic performance and reliability are high.

关 键 词：有载分接开关 变压器绕组 故障自动化诊断 LSTM 欧氏距离 

分 类 号：TM71[电气工程—电力系统及自动化]