Improved Combinatorial Methods For Reliability Analysis Of Competing Failures System

Reliability analysis is an important task in the design,tuning and maintenance of complex systems.Accurate reliability analysis can help effectively avoid risks and reduce losses.This paper studies the dynamic system reliability with competitive failure behavior.Competitive failure behavior refers to the time-domain competitive relationship between the local failure of the trigger component and the propagated failure of the dependent component in the function dependent system: the propagated failure of the dependent component occurs first,and other system components are affected and fail;the local failure of the trigger component occurs first,the dependent component is isolated,and other system components are not affected.In dealing with this kind of dynamic competitive system,the combinatorial method is usually used.The combinatorial method adopts the idea of divide and conquer,divides the system into several static subsystems for analysis and evaluation.The existing combinatorial methods for static system analysis are based on fault tree reduction.This kind of methods need to generate multiple reduced fault trees and analyze the fault trees for many times,which is inefficient.In this paper,two improved combinatorial methods based on binary decision diagram(BDD)are proposed: method one is called BDD reduction combinatorial method,which generates BDD according to system fault tree,then generates BDD according to different events,then evaluates the conditional failure probability of the reduced BDD,and finally calculates the system's unreliability;method two is called BDD assignment combinatorial method which is to directly evaluate BDD to get the calculation formula of the conditional failure probability of the system according to the system fault tree and BDD node's edge probability according to the conditional failure probability of components under different event occurrence conditions,then substitute the edge probability into the calculation formula to get the conditional failure probability of the system and finally calculate the system's unreliability.In this paper,these two methods are applied in the reliability analysis of two case systems,the practicability of the method is proved,and the correctness of the method is verified by using Markov method.Finally,the efficiency of the method is compared with the combinatorial method based on fault tree reduction.The results show that the improved combinatorial method can reducethe numbers of fault tree generation and the times of transformation from fault tree to BDD model,and the analysis efficiency is higher.