矩阵加权关联规则在故障诊断系统中的应用  

Application of Matrix-Weighted Association Rule Mining Algorithm to Fault Diagnosis

作　　者：朱清香[1] 焦朋沙[1] 刘晶[2] 郝红红[1] 

机构地区：[1]燕山大学经济管理学院,河北秦皇岛066004 [2]燕山大学图书馆,河北秦皇岛066004

出　　处：《工业工程》2013年第2期87-91,96,共6页Industrial Engineering Journal

基　　金：河北省自然科学基金资助项目(G2010001331)

摘　　要：关联规则挖掘算法实现了对复杂设备的通用、快速、脱离主观经验的故障诊断。经典的关联规则算法以各项目均匀分布为前提,而实际的故障诊断过程中不同的故障因素对故障诊断的贡献度不同。针对这种情况,将"最小支持期望"和矩阵引入关联规则,提出一种适用于设备故障诊断的基于矩阵的加权关联规则模型——MWARMA模型,实例证明该模型在提高挖掘效率的同时,明显提高了故障诊断的准确率。以该模型为基础设计并实现了一套设备故障诊断系统。By the association rule mining algorithm, it can diagnose fauhs of complex equipment in a gen- eral and fast way without the need of subjective experience. The drawback is that the classical association rule algorithm requires that the frequency and importance of the items should be similar. However, in practical fault diagnosis applications, the contribution of each fault factor is different. To solve this prob- lem, a new model called matrix-based weighted association rule mining algorithm suitable for equipment fault diagnosis is proposed by introducing min-support expectation. Experiments show that the model im- proves the diagnostic efficiency while obviously increasing the accuracy of fault diagnosis. Then, an equip- ment fault diagnosis system is designed and implemented based on matrix-base weighted association rule mining algorithm （MWARMA） model.

关 键 词：故障诊断 专家系统 加权关联规则 最小支持期望 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]