基于EFPI传感器的GIS局部放电模式识别研究  

Research on GIS partial discharge pattern recognition based on EFPI sensor

作　　者：何彦良 尚宇 任双赞 杨昌建 牛博 HE Yanliang;SHANG Yu;REN Shuangzan;YANG Changjian;NIU Bo(Power Institute of State Grid Shannxi Electric Power Company Limited,Xi’an 710100,China)

机构地区：[1]国网陕西省电力有限公司电力科学研究院,陕西西安710100

出　　处：《传感器与微系统》2024年第7期18-21,共4页Transducer and Microsystem Technologies

摘　　要：为了在复杂电磁环境中精准捕捉局部放电产生的微弱信号,设计了基于非本征法布里—帕罗干涉(EFPI)传感器的GIS局部放电模式识别方法。通过差分双波长光强强度比解调算法,粗解调EFPI传感器的腔长;利用复域相关算法,结合粗解调结果,精解调腔长,确定最终的EFPI传感器腔长,用于检测GIS设备的超声波信号;通过希尔伯特(Hilbert)变换GIS设备的超声波信号,得到Hilbert边际谱;利用2维—1维深度残差网络,提取Hilbert边际谱的特征,输出GIS局部放电模式识别结果。实验分析结果表明:该方法可有效确定EFPI传感器的腔长,采集GIS设备的超声波信号;得到GIS设备超声波信号的Hilbert边际谱;在复杂噪声环境下,该方法可精准识别GIS局部放电模式。In order to accurately capture weak signals generated by partial discharge in complex electromagnetic environment,gas-insulated switchgear(GIS)partial discharge pattern recognition method based on EFPI sensor is designed.The cavity length of EFPI sensor is coarse demodulated by differential dual-wavelength light intensity ratio demodulation algorithm.Using complex domain correlation algorithm,combined with coarse demodulation results and precise demodulation cavity length,the final EFPI sensor cavity length is determined,which is used to detect ultrasonic signals of GIS equipment.Hilbert marginal spectrum is obtained by transforming ultrasonic signal of GIS equipment by Hilbert.The features of Hilbert marginal spectrum are extracted by using 2D—1D depth residual network,and the results of partial discharge pattern recognition in GIS are output.Experimental analysis result shows that:this method can effectively determine the cavity length of EFPI sensor and collect ultrasonic wave signal of GIS equipment;obtain the Hilbert marginal spectrum of ultrasonic signals of GIS equipment.Under complex noise environment,this method can accurately identify GIS partial discharge mode.

关 键 词：非本征法布里—帕罗干涉传感器 气体绝缘开关设备 局部放电 模式识别 复域相关 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] TM855[自动化与计算机技术—计算机科学与技术]