基于改进分层有向图的故障诊断方法

A fault diagnosis method based on improved hierarchical directed graph

针对火电机组热力系统内部因果关系复杂、故障诊断系统实时性差以及分辨率低等问题,提出了基于符号有向图(SDG)的分层有向图模型,以及基于改进分层有向图的故障诊断方法,即采用有向图分层策略,使故障源的搜索空间减小;利用故障报警节点回溯路径上的定性关系,搜索相容支路确定故障源候选节点。将分层有向图故障诊断方法应用于某台除氧器的故障诊断,并与传统的SDG故障断方法进行比较,结果表明,分层有向图故障诊断方法的有效节点搜索空间明显小于传统的SDG方法,从而提高了系统的诊断速度;使故障源节点的分辨率达到100%,满足实时性要求。

To solve problems like complex internal causality of large-scale thermal systems,poor real-time,low resolution ratio of fault diagnosis systems and other ones,the formal definition of hierarchical orientation graph model based on signed directed graph （SDG） was proposed.Moreover,a fault diagnosis method was also put forward,on the basis of the improved hierarchical directed graph.In this method,a directed graph segmentation strategy is applied to reduce the search space for failure source;and the qualitative relationship on the backward path of failure alarm node is used to search for compatible branch to determine the source of the fault candidate nodes.Furthermore,this method was applied on a deaerator system for fault diagnosis,and the results were compared with that obtained by the conventional SDG fault diagnosis method.The results show that,search space of this hierarchical orientation graph diagnosis method is dramatically smaller than that of the conventional method,indicating this new method can improve the diagnostic system speed,make the resolution ratio of the source of fault nodes reach 100％,which meets requirements of real-time.