基于软打分机制—多模态融合的真实开关柜局部放电模式识别方法  

Partial Discharge Pattern Recognition Method for Real Switch Cabinet Based on Soft Scoring Mechanism and Multi⁃mode Fusion

作　　者：刘树慰[1] 邸振国 李斯盟 卜泽伟 徐建东[1] 汲胜昌[2] LIU Shuwei;DI Zhenguo;LI Simeng;BU Zewei;XU Jiandong;JI Shengchang(Offshore Oil Engineering Co.,Ltd.,Tianjin 300451,China;State Key Laboratory of Electrical Insulation and Power Equipment,Xi’an Jiaotong University,Xi’an 710049,China)

机构地区：[1]海洋石油工程股份有限公司,天津300451 [2]西安交通大学电力设备电气绝缘国家重点实验室,西安710049

出　　处：《高压电器》2023年第8期203-211,共9页High Voltage Apparatus

基　　金：国家自然科学基金项目(52007148);强脉冲辐射环境模拟与效应国家重点实验室开放基金项目(SKLIPR2114)。

摘　　要：局部放电类型与开关柜早期绝缘劣化程度密切相关,准确识别开关柜局部放电类型是改善开关柜运行状态,保障电网安全的重要手段。文中设计了真实开关柜中易产生的尖端放电、悬浮电位放电、强垂直分量沿面和弱垂直分量沿面放电4种局部放电类型。利用特高频传感器、暂态地电波传感器、超声传感器和高频电流线圈对开关柜的局部放电进行多模态融合检测,分析了同一缺陷下不同传感器的检测灵敏度以及不同缺陷类型的谱图特性。提取谱图的放电密度分布特征参量为正负峰值差Dpn、边沿积分差值Dei、前端均值差Dav、前端放电数占比Pfd和相位横向跨度Tr。提取谱图的轮廓特征参量为与缺陷标准轮廓的相关系数r。将所提取的特征参量融合得到表征开关柜局部放电模式的新特征体系,并提出了一种基于软打分机制的开关柜局部放电模式识别方法。该方法解决了传统识别方法谱图特征单一、识别准确率低、识别类型不精准的问题,为开关柜设备的在线监测和故障诊断提供了理论依据。Type of partial discharge is closely related to the early insulation deterioration degree of switch cabinet.Identifying the type of partial discharge of switch cabinet accurately is an important means to improve the operating condition of the switch cabinet and ensure the safety of power grid.In this paper,such four kinds of partial discharge as tip discharge,suspension potential discharge,strong vertical component along the surface discharge and weak ver-tical component along the surface discharge are designed.The multi-mode fusion detection of partial discharge of the switch cabinet is carried out by using UHF sensor,TEV sensor,AE sensor and HFCT,and the detection sensitivity of different sensors and the spectral characteristics of different defect types under the same defect are analyzed.The characteristic parameters of discharge density distribution in extraction spectra are peak difference between positive and negative Dpn,edge integral difference Dei,front-end mean difference Dav,front-end mean difference,Pfd and phase transverse span Tr.The contour characteristic parameter of the extraction spectrum is the correlation coefficient r with the defect standard contour.A new feature system is obtained by fusion of the extracted characteristic parame-ters,and a soft scoring method is proposed to identify the partial discharge mode of switch cabinet.This method solves the problems of single spectral feature,low recognition accuracy and inaccurate recognition type of traditional recognition methods,and provides a theoretical basis for online monitoring and fault diagnosis of switch cabinet equipment.

关 键 词：真实开关柜 局部放电 多模态融合 软打分机制 模式识别 

分 类 号：TM591[电气工程—电器] TM855