Research On Reliability Analysis Method Of Phase Mission System Considering Human Error

Many complex systems in the fields of aerospace,rail transit,communication network and so on belong to Phased Mission Systems(PMS),whose reliability has a significant impact on the safe operation of these systems.Studying their reliability is conducive to improving system design and developing appropriate maintenance schemes.Therefore,this paper proposes a reliability analysis method for PMS’s fault characteristics based on discrete-time Bayesian network,which considers human error factors.The main research focuses on the following parts:Firstly,aiming at the construction of the PMS fault model,the paper utilizes a dynamic fault tree to analyze the fault mechanisms and construct models for individual phase.These models are then integrated into the entire mission system fault model considering the correlations between different phases.To address the issues of epistemic uncertainty and multiple life distributions in PMS,some methods are proposed in this paper.Failure rates of components subject to exponential distribution are obtained by integrating the evaluation values of multiple experts,and a coefficient of variation method is used to estimate distribution parameters of components subject to the Weibull distribution.Secondly,redundancy techniques are widely used in modern systems,and the failure of a single component has a little impact on the system.Instead,system failures caused by related failures are gradually increasing.This paper considers two types of related failures: common cause failures and competing failures.For common cause failures,the paper regards them as basic events to construct a discrete-time Bayesian network model and uses a β-factor model to solve the common cause failure rate.For competing failures,the paper analyzes logical relationships between the local failures of triggering components and the propagation failures of related components in a competing system,and maps them into a discrete-time Bayesian network for analysis.Additionally,some reliability parameters can be obtained based on the inference algorithm of discrete-time Bayesian networks.Finally,a case study is used to verify the importance of common cause failures and competing failures in reliability analysis.Additionally,with regard to the issue of human error in PMS reliability analysis,this paper proposes a human reliability analysis method based on a discrete-time Bayesian network and SPAR-H method.Each performance shaping factor(PSF)in the SPAR-H method is rated,and its weight factor is determined.A human reliability analysis model is established based on the phase dependence of PMS,and the conditional probability distribution tables of each node in the model are determined to achieve quantitative calculation.Finally,the importance of human error in reliability analysis is demonstrated through a case study.Lastly,reliability results and risk priority number are used to construct a multiattribute decision matrix,and an improved multi-attribute decision algorithm is proposed based on evaluation of distance from average solution(EDAS),which uses the combination weighting method to weight different attributes.The important maintenance components are determined,and the maintenance strategy is developed by sorting the evaluation scores of each scheme.Finally,taking the process of an aircraft performing a flight mission as the research object,a reliability analysis method of phased mission systems is proposed based on discrete-time Bayesian networks.The analysis results can be used to develop the appropriate maintenance strategy for systems,and the feasibility of the proposed method for reliability analysis of PMS is demonstrated.