基于故障树分析的多位置多尺寸混合故障的智能诊断系统

Intelligent Diagnosis System of Multi Position and Multi Size Mixed Fault Based on Fault Tree Analysis

针对旋转机械系统中,故障因素复杂多样,且复合故障发生概率大于单一故障等问题,文章借助VB.NET提出一种基于故障树分析的多位置多尺寸混合故障的智能诊断系统。该智能诊断系统分为规则推理和案例匹配两大部分。通过故障树的分析获取故障信息,而且故障树的每个节点对应规则库中若干条诊断规则,结合K-Means、FCM、SOM等故障聚类方法,得到轴承内圈、外圈和滚动体故障的敏感特征及其相应的诊断规则,并制定基于LMD分解的诊断规则,实现对轴承的9种混合故障进行故障位置、尺寸和程度的详细诊断;建立故障案例的案例库,以故障特征和特征值为征兆信息将问题案例存储到案例库中。在进行案例匹配时先对案例库进行分类检索,缩小匹配范围,以提高检索效率。在进行规则诊断时运用特征值判定算法获得其诊断向量,并通过各规则属性的权重以及规则自身的置信度,解决规则诊断时可能出现的推理冲突问题,实现准确快速地故障识别。

In this paper,an intelligent diagnosis system based on Fault Tree Analysis Based on fault tree analysis is proposed in this paper to solve the problems of complex and diverse fault factors in rotating machinery system and the probability of complex fault occurrence is greater than single fault.The intelligent diagnosis system is divided into two parts:rule reasoning and case matching.Through the analysis of fault tree,the fault information is obtained,and each node of the fault tree corresponds to some rules in the rule base.Combining the K-Means,FCM,SOM and other fault clustering methods,the sensitive characteristics of the fault of the inner ring,outer ring and rolling body of the bearing and the corresponding diagnosis rules are obtained,and the diagnosis rules based on the LMD decomposition are set up.A detailed diagnosis of the fault location,size and degree of the 9 kinds of hybrid faults is carried out,and the case base of the fault case is set up,and the problem cases are stored in the case base with the fault features and the eigenvalues as the sign information.In case matching,the case database is classified and retrieved,and the matching scope is narrowed,so as to improve retrieval efficiency.In the process of rule diagnosis,the feature value decision algorithm is used to obtain the diagnosis vector,and through the weight of the rules of the rules and the confidence of the rule itself,the reasoning conflict problems may appear in the rule diagnosis,so that the fault recognition can be realized accurately and quickly.