Research On Fault Diagnosis Of Causality Diagram Based On Binary Decision Graph And Fuzzy Inference

In the field of fault diagnosis of complex systems,the uncertainty problems caused by the complexity of objects,the limitation of testing methods,the imprecision of knowledge and so on are the majority,which makes the fault diagnosis problems complicated.Dynamic causality diagram originated from reliability network is a tool that applies probability statistics to the complex fields and the expression and reasoning of uncertain knowledge.The simple knowledge expression and the flexible reasoning methods make the causality diagram more and more widely used in the field of fault diagnosis.This paper focuses on the knowledge expression and reasoning of the causality diagram.The main contents are as follows:(1)Aiming at the problem that the fault diagnosis method of the minimum cut set of causality diagram needs to sort the minimum cut set,this paper proposes a new reasoning method.Firstly,the causality diagram is transformed into a Binary Decision Diagram(BDD),the disjoint cut set is written by the directly search path,and then the minimum cut set is obtained by using the Boolean Absorption Principle.Secondly,according to the similarity between BDD structure and Huffman tree,the minimum cut set is coded by Huffman coding and the length of Huffman coding is regarded as the structural importance of the minimum cut set,and the minimum cut set is grouped and sorted accordingly,so as to simplify the sorting process.(2)In view of the uncertainty of event occurrence probability and the complexity of causality between events,this paper introduces fuzzy number into causality diagram to describe the uncertainty information in the real world.Firstly,we use the triangle fuzzy number to replace the imprecise probability of the event and use the minimum cut set to solve the probability of intermediate events.Secondly,according to the many-to-many and intercrossing phenomena of fault,we combine the causality diagram with fuzzy reasoning to define the fuzzy relation matrix,and then use the matrix operation to solve the most probable reason of fault.(3)Based on the superiority of reliability network model for the polymorphism expression of event,this paper directly applies the joint distribution probability and conditional probability formula of the reliability network to the multi-valued causality diagram inference.The example proves that the proposed algorithm is reasonable.