基于深度神经网络的超高压开关故障智能检测方法  

Deep Neural Network-based Intelligent Fault Detection for Ultra-high Voltage Switches

作　　者：王丹 高治良 林东跃 WANG Dan;GAO Zhiliang;LIN Dongyue(Shenzhen Micro&Nano Perception Computing Technology Co.,Ltd.,Shenzhen 518100,China)

机构地区：[1]深圳市微纳感知计算技术有限公司,广东深圳518100

出　　处：《电工技术》2025年第1期191-196,208,共7页Electric Engineering

基　　金：中国南方电网有限责任公司创新项目“音频矩阵感知技术在开关智能巡视的应用研究”(编号CGYKJXM20220117)。

摘　　要：针对超高压开关故障定位和异常预测问题,提出了一种基于深度神经网络的超高压开关故障检测方法。该方法利用多种传感器和数据源,自动提取并学习超高压开关设备的特征,实现超高压开关故障的分类和定位。同时,该方法还可以对超高压开关进行异常预测,提前发现潜在故障隐患,为超高压开关设备的维护和管理提供支持。实验结果表明该方法在超高压开关故障定位和异常预测方面具有很高的准确性和鲁棒性。Aiming at fault location and anomaly prediction of ultrahigh voltage switches(UHV),this study made a beneficial attempt at establishing a deep neural network-based method of detecting the faults.The method was designed to use a variety of sensors and data sources to automatically extract and learn the characteristics of the equipment,and thereby to realize the classification and location of UHV switch faults.Furthermore,the method was also expected to predict anomalies of UHV switches and discover the potential faults in advance,so as to help in maintenance and management of UHV equipment.The proposed method was indicated by experiments highly accurate and robust in locating faults and prognosing anomalies of UHV switches.

关 键 词：深度神经网络 超高压开关 故障定位 故障分类 异常预测 

分 类 号：TM76[电气工程—电力系统及自动化]