基于朴素贝叶斯的局部放电诊断模型  

A PARTIAL DISCHARGE DIAGNOSIS MODEL BASED ON NAVE BAYES

作　　者：陈新美 潘笑颜 路光辉[3] 牧继清[3] 姬波[2] 

机构地区：[1]许昌开普检测技术有限公司,河南许昌461000 [2]郑州大学信息工程学院,河南郑州450002 [3]许继集团有限公司,河南许昌461000

出　　处：《计算机应用与软件》2016年第9期51-55,共5页Computer Applications and Software

基　　金：国家自然科学基金项目(61170223);河南人才培养联合基金项目(U1204610);河南省科技攻关计划项目(132102210404)

摘　　要：针对局部放电故障诊断问题,提出一种基于朴素贝叶斯的局部放电诊断模型,并对模型中的朴素贝叶斯的应用方法进行详细研究。该模型由四部分组成:信号的接收及处理、谱图产生、特征提取和朴素贝叶斯分类。诊断流程:首先由UHF传感器接收局部放电信号并交于信号调理单元处理;然后基于处理后的信号产生三维谱图,提取谱图的典型特征;最后采用朴素贝叶斯算法进行故障诊断。该模型已作为插件嵌入到某一电力设备生产企业的变压器监测产品中。实际测试表明该模型较好地满足了应用需求。For the problem of partial discharge fault diagnosis, we present a Nai＇ve Bayes-based partial discharge diagnosis model, and study in detail the application method of Naive Bayes in the model. The model consists of four components： reception and processing of signals, spectrum maps generation, feature extraction and Naive Bayes classification. The diagnosis flow is as follows： first the UHF sensor is used to receive partial discharging signals and transmits them to signal conditioning unit for processing. Then based on the processed signals the model generates three-dimensional spectrum map, and extracts typical features of the map. Finally it uses Naive Bayesian algorithm to diagnose the faults. The model as an important plug-in has been embedded into the transformer monitoring products of an enterprise of power equipment production. Practical tests show that the model well satisfies the application requirements.

关 键 词：局部放电 朴素贝叶斯 诊断模型 特征提取 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]