| Fault tree analysis is the important method for reliability analysis.The traditional fault tree use static logic gates such as AND or OR to describe event relationships,which cannot express the dynamic failure characteristics in the hydraulic system.Dynamic fault tree analysis has the higher capability in describing the dynamic failure behavior existing in the hydraulic system.Dynamic fault tree is often solved by discrete-time Bayesian network,whose computing time and calculation error are affected by the time segmentation.Therefore,an improved method for solving dynamic fault tree based on discrete-time Bayesian network is proposed to decrease the computing time and error.Fault tree analysis method is often on the premise that the failure of the basic event is independent.This premise is inconsistent with the fact and causes errors.To this end,a DFTA method considering dependent failures and its solution method based on discrete-time Bayesian network is proposed.First,the analysis method of static and dynamic failure behavior in hydraulic system based on dynamic fault tree is studied.The event relationships described by the logic gates of dynamic fault tree and their performance in hydraulic systems are studied.The acquisition methods of hydraulic components' reliability index and the method for its correcting is proposed.Secondly,in accordance with these problems that the calculation efficiency and accuracy of the method for solving dynamic fault tree based on discrete-time Bayesian network are affected by the time segmentation,an improved method is proposed.First,an improved method of transforming the dynamic fault tree to discrete-time Bayesian network is used to make the conditional probability table more accurate.Then,a modular improvement method of discrete-time Bayesian network is proposed,which reduces the scale of the conditional probability table.Finally,the calculation efficiency and accuracy are improved.Then,in view of the issue that the case which assumes the basic events are independent in fault tree method will bring calculation errors,a dynamic fault tree method considering dependent failures and its solution based on discrete-time Bayesian network is proposed.For different applications,a conditional probability model based on expert experience and a joint probability model based on the Copula model are proposed to describe the dependent failure features between components.Finally,the dynamic fault tree method considering dependent failures is used to analysis the reliability of the dumper hydraulic system,and the improved method for solving dynamic fault tree based on discrete-time Bayesian network is used to solve it.The failure probability of system and the posterior probability,Birnbaum importance measure,and criticality importance measure of the components are obtained to judge the contribution and impact of the components on the system from different perspectives. |
PDF Full Text Request