基于告警信息和CNN的变电站故障诊断方法研究  

Research on Substation Fault Diagnosis Method Based on Alarm Information and CNN

作　　者：李波 游剑铭 刘晓放 王荣 冷贵峰 马建伟 Li Bo;You Jianming;Liu Xiaofang;Wang Rong;Leng Guifeng;Ma Jianwei(Xingyi Power Supply Bureau,Guizhou Power Grid Co.,Ltd.,Xingyi 562400,China;Power Dispatching Control Center,Guizhou Power Grid Co.,Ltd.,Guiyang 550000,China)

机构地区：[1]贵州电网有限责任公司兴义供电局,贵州兴义562400 [2]贵州电网有限责任公司电力调度控制中心,贵阳550000

出　　处：《煤矿机械》2024年第4期161-164,共4页Coal Mine Machinery

摘　　要：智能变电站告警信息包含丰富的设备状态信息。根据告警信息文本的特点,提出了一种基于告警信息和卷积神经网络(CNN)的变电站故障诊断方法。首先通过分析大量变电站告警信息,总结出告警信息文本的类别和特点;然后根据中文文本分类方法和告警信息文本特点,建立了基于CNN的变电站故障诊断模型;最后通过算例对基于CNN的变电站故障诊断模型进行训练分析。结果表明,相比于传统机器学习分类模型,该方法能够明显提高变电站故障诊断准确性。The alarm information of intelligent substations contains rich equipment status information.A substation fault diagnosis method based on alarm information and convolutional neural network(CNN)was proposed according to the characteristics of alarm information text.Firstly by analyzing a large amount of substation alarm information,the characteristics of the alarm information text are summarized.Then according to the Chinese text classification method and the characteristics of alarm information text,a substation fault diagnosis model based on CNN was established.Finally an calculation example is used to train and analyze the substation fault diagnosis model based on CNN.The results show that compared to traditional machine learning classification models,this method can significantly improve the accuracy of substation fault diagnosis.

关 键 词：变电站 告警信息 故障诊断 CNN 

分 类 号：TM417[电气工程—电器]