The Fault Location Method Based On Hierarchical Strategy And Fuzzy Five-range SDG

With the development of science and technology, the complexity of thechemical process has improved. The chemical raw material is inflammable andexplosive, and the chemical production environment is generally in extremeconditions. Thus, the fault will cause big damage. The method of how to quicklyand efficiently reason and find the source of fault has become a hot topic in thefield of chemical safety. The signed directed graph (SDG) is widely used inindustrial fault diagnosis and fault location due to its good informationcompleteness, and it can quickly describe the fault propagation process withoutthe accurate mathematical model.The qualitative information of traditional SDG cannot describe the faultseverity, and it needs a lot of time to reason the fault of complex system. Thetraditional SDG can not further distinguish fault source from multiple candidates.The existed SDG diagnosis methods improve the traditional SDG from a certainangle and lead to the emergence of some problems: the diagnosis efficiencytowards complex system will decrease after the expansion of model information;the model ignores the dynamic information of fault and cannot pinpoint thesource of fault, etc. In order to solve these problems, a comprehensive fault diagnosis method is proposed from the aspects of structure, model and reason inorder to improve SDG model.In traditional SDG model, the quantitative information of model isinsufficient, and the resolution of the fault is low. The variable node has onlythree states, and the fault information is poor. The node state is establishedaccording to the threshold, and the setting of threshold is the important criteriato determine whether the fault occurs. The traditional SDG model is establishedby the setting of single threshold. The node can only describe the occurrence offault, but cannot reflect the severity. In addition, the clear value cannotaccurately describe the fault information due to the existence of noise in themeasurement process, and a precise way is needed to describe the systemvariables. In order to remedy the shortcomings of low resolution and insufficientquantitative information of SDG model, the three-range of node state isextended to five-range. Moreover, the fuzzy set theory is introduced in the paperto establish the fuzzy five-range SDG.The expansion of model information will increase the complexity of systemstructure, and the search of fault becomes more difficult accordingly. Thestructure of the SDG model can be simplified based on hierarchical strategy. Inaddition, the fault can only propagate from high hierarchy to low hierarchy inhierarchal SDG model. Thus, the fault candidates can be obtained quickly andthe search space is reduced greatly.The fault search method always uses backward tracking approach, but it cannot further search the source of fault due to insuffcient diagnosticinformation. In the paper, the fault location algorithm is introduced from theperspective of reasoning. The localization problem of fault source can beconverted to the sorting problem of consistent path priority based on the settingof common point. Moreover, the fault magnification, fault support degree andfault velocity are introduced to locate the fault source. It enhances the precisionof fault location and remedies the shortcomings that traditional SDG onlycontains static information. Finally，the TEP is used as an example for faultsimulation analysis to verify the effectiveness of the method. The simulationresult shows that the method is efficient and accurate. The method has goodinformation completeness and an excellent ability of fault location, and it plays arole in reducing equipment loss and improving the safety of chemical industrysystem.