A Novel Continuous-time Bayesian Network Analysis Method

Bayesian network need to be constructed based on fault tree.In static systems,static Bayesian network and static fault tree are combined to expand the method of solving fault tree,but it can only describe static failure behaviors and cannot describe dynamic failure behaviors such as time dependency,functional dependency,sequential,and redundancy;in dynamic systems,discrete-time and continuous-time Bayesian networks are combined with Dugan dynamic fault trees to simplify the calculation of Dugan dynamic fault trees,but it is difficult to describe all the static and dynamic failure behaviors of the system,so the existing discrete-time and continuous-time Bayesian network analysis methods based on Dugan dynamic fault tree have limitations.T-S dynamic fault tree analysis method breaks through the limitations of Dugan dynamic fault tree analysis method,can characterize any static and dynamic failure behaviors.In view of the above shortcomings,in order to give full play to the advantages of T-S dynamic fault tree and Bayesian network in analysis modeling and reasoning calculation,a novel continuous-time Bayesian network analysis method is proposed,which is continuous-time Bayesian network analysis method based on T-S dynamic fault tree.And the importance and sensitivity analysis methods based on this method are proposed.Firstly,a novel continuous-time Bayesian network analysis method is proposed.The method of converting T-S dynamic fault tree to continuous-time Bayesian network is given.The algorithms of leaf node failure probability and root node posterior probability of continuous-time Bayesian network are proposed.In order to verify the feasibility of the proposed method,it is compared with static Bayesian network analysis method,discrete-time and continuous-time Bayesian network analysis methods based on Dugan dynamic fault tree,discret-time and continuous-time T-S dynamic fault tree analysis methods.Secondly,Birnbaum importance measure(probability importance measure),criticality importance measure,differential importance measure,integrated importance measure and sensitivity of novel continuous-time Bayesian network are proposed,and static,discrete-time Bayesian network importance measure and sensitivity algorithms are given.The importance measure and sensitivity algorithms of the proposed method are compared by the importance measure and sensitivity algorithms of static,discrete-time Bayesian network,and the feasibility of the importance measure and sensitivity algorithms of the continuous-time Bayesian network is verified.Finally,the proposed novel continuous-time Bayesian network method is applied to the reliability analysis of the hydraulic break system of the wind turbine and satellite power system.The system's failure probability and each component's posterior probability,Birnbaum importance measure,criticality importance measure,differential importance measure,integrated importance measure,upgrading function,risk performance value,risk reduction value and sensitivity of each component are solved,and the results are analyzed to provide a basis for helping find weaknesses in the system,improving system design,fault diagnosis and maintenance.