基于事件序列数据挖掘原理的高压输电线系统故障诊断(一)模型与算法  

A NEW DATA MINING APPROACH FOR FAULT DIAGNOSIS OF HV TRANSMISSION LINE SYSTEM BASED ON SEQUENCE OF EVENTS Part One Model and Method

作　　者：廖志伟[1] 孙雅明[1] 

机构地区：[1]天津大学电气与自动化学院

出　　处：《电力系统自动化》2004年第4期22-27,共6页Automation of Electric Power Systems

基　　金：国家自然科学基金资助项目(59877016)

摘　　要：基于事件序列(SOE)数据挖掘(DM)原理的高压输电线系统故障诊断的研究分两部分,该文是第一部分:模型与算法;第二部分是仿真和容错性能分析。根据电力系统发生故障时形成具有时空特征的离散SOE信息,提出了基于事件序列DM原理的故障诊断模型的构造和实现方法。用动态规划算法的相似性优化原理,对实时信息序列中含有畸变信息的SOE与标准的SOE集合之间进行编辑距离运算、运算代价、置信度来完成相似性度量的操作,可由挖掘事件之间的蕴涵的关联性,发现知识和进行信息纠错。所提出方法的重要实质是把故障诊断问题的求解转化为寻找与实时故障SOE模式最相似的、运算操作代价最小的标准故障SOE模式,从而纠正SOE中的变异事件,确保诊断的高容错性能。This is a 2-part paper. A sequence of events (SOE) based data mining (DM) model and method for fault diagnosis in power system is presented in part 1. The emphasis of this part is to present the basic definition of sequence of events and the essential theory and associated realization method. According to the time-space characteristics of discrete SOE information in power systems, this paper proposes a new fault diagnosis model of SOE based DM principle and the realization. Furthermore, dynamic programming algorithm is used for editing operation to find the similarity between event sequences in presented model. The similarity of event sequences is measured by editing distance operation, cost of operation and confidences of event. Then fault diagnosis problem is converted into seeking the most similar normal event sequence to real time fault event sequence with the lowest operation cost. Henceproposed DM model has a high fault-tolerance performance to correct variant information in SOE of fault diagnosis.

关 键 词：输电线系统 故障诊断 数据挖掘 事件序列 动态规划法 容错性能 

分 类 号：TM769[电气工程—电力系统及自动化]