基于视觉检测的间歇性接地故障研究  

Research on Intermittent Grounding Fault Based on Visual Detection

作　　者：姜文斌 苏君 柴俊 陆增洁 JIANG Wen-bin;SU Jun;CHAI Jun;LU Zeng-jie(State Grid Shanghai Electric Power Co.,Ltd.Shibei Power Supply Company,Shanghai 200072,China)

机构地区：[1]国网上海市电力公司市北供电公司,上海200072

出　　处：《电工电气》2022年第6期8-13,共6页Electrotechnics Electric

基　　金：国网上海市电力有限公司科学技术项目(SGTYHT/20-JS-224)。

摘　　要：线路发生间歇性接地故障时接地电流较小,并且具有瞬间性的特点。针对瞬时接地电流会导致接地电阻温度升高的特点,提出对主变接地电阻等装置以200ms的采样频率进行非接触视觉检测以及监控瞬时温升,获取接地电阻的表面、室内外、油枕油位等温度数据,利用检测得到的大量数据通过BP神经网络来优化诊断结果,从而改善系统检测的滞后性,达到快速判断并处理故障。基于数据驱动的数学模型,对间歇性接地故障进行预警和告警,并通过实验验证了该方案的可行性与有效性。The grounding current is small when the line meets intermittent grounding fault,and it has an instantaneous characteristic.The instantaneous grounding current could trigger the temperature rise of the grounding resistance.This paper proposed non-contact visual detection and instantaneous temperature rise monitor for the sampling frequency of 200 ms of the grounding resistance device of the main transformer to collect temperature data in different locations.This paper used the detected data and the BP neural network to optimize the diagnostic result.It improved the hysteresis of the system and helped to detect and manage the faults.In addition,it could warn and alarm the intermittent grounding fault based on the data-driven model.The experiment verifies the feasibility and effectiveness of the scheme.

关 键 词：间歇性接地故障 视觉检测 红外热成像 边缘计算 数据驱动模型 

分 类 号：TM862[电气工程—高电压与绝缘技术]