基于事件序列数据挖掘原理的高压输电线系统故障诊断 (二)仿真和容错性能分析

A NEW DATA MINING APPROACH FOR FAULT DIAGNOSIS OFHV TRANSMISSION LINE SYSTEM BASED ON SEQUENCE OF EVENTSPart Two Simulation and Fault-tolerance Performance Analysis

在基于事件序列(SOE)数据挖掘原理的故障诊断模型与方法的基础上,提出了将基于事件序列的数据挖掘原理故障诊断模型用于高压输电线系统(HVTLS)故障诊断领域。分析了故障诊断模型中变异故障模式的种类及其对故障诊断策略的影响,阐述了基于SOE的数据挖掘HVTLS诊断模型和功能模块的构造。依据输电系统故障事件序列在时空特性的关联性,用动态规划算法的优化相似性原理挖掘事件序列之间的关联性和蕴涵的知识,将诊断问题的求解转化为寻找与实时故障事件序列模式最相似的、运算操作代价最小的标准故障事件序列模式。以实现对变异事件序列模式中畸变事件的“复原”和纠错,从而确保故障诊断系统的高容错性。

Based on the sequence of events (SOE) based data mining (DM) fault diagnosis model for the fault diagnosis of high-voltage transmission line system (HVTLS), the model and the functions are described in this part. The classification and influence of variation fault patterns (VFP) derived from distorted information are analyzed respectively for diagnostic policy. According to the time-space characteristic and associated knowledge between events in SOE, dynamic programming algorithm is used for mining associated knowledge in the presented DM model. Therefore, fault diagnosis problem is converted into seeking the most similar standard fault patterns (SFP). The simulation results show that VFP is restored to their antitype fault patterns by using the spatial-temporal characteristics of SOE alarm information. The higher fault-tolerant performance of fault diagnosis system is guaranteed in presented approach. This work is supported by National Natural Science Foundation of China ( No. 59877016).