Reliability Evaluation And Selective Maintenance For Complex Systems With Inspection Data

With advanced manufacturing systems and military equipment designed towards lager scale,more complex and intelligent,it is a challenging task to ensure their reliability in operational stages as they may possess highly integrated functions,improved performances,and complicated loads,and be exposed to severe environments.To overcome this problem,it is crucial to track the degradation trend of these systems,assess health status,predict future reliability,and make effective maintenance planning.On the one hand,as engineered systems may exhibit multi-state nature during their failure processes,modeling the failure mechanism and deterioration process of multi-state systems has received considerable attentions in recent years.On the other hand,some engineered systems are required to perform a sequence of missions and maintenance activities can be performed between two consecutive missions.Optimally selecting a subset of feasible maintenance actions for components under limited maintenance resources can improve systems' operational reliability effectively.With the development of advanced sensors and condition monitoring tools,inspection data reflecting systems' health status can be collected in the systems' operational stage.Utilizing the inspection data can effectively track and predict the trend of deterioration and faciliate maintenance planning.Additionally,maintenance decision-making needs to consider allocating limited maintenance resources among multiple missions.This dissertation devotes to infer the unknown state transition parameters for multi-state systems by aggregating multi-level inspection data and explore selective maintenance strategies for complex systems under imperfect inspections.The primary research contributions and innovative outcomes are summarized as follows:(1)Development of a parameter inference method for multi-state systems by aggregating multi-level inspection sequences.The proposed inference method generally consists of two stages: firstly,compute the sequences of the posterior state probability distributions of units based on multi-level inspection sequences by dynamic Bayesian network models;secondly,estimate the unknown transition probabilities of units by converting the sequences of posterior state probability distributions into a least squares problem.Two illustrative examples,together with a set of comparative studies,are presented to demonstrate the effectiveness and efficiency of the proposed method.(2)Development of a robust selective maintenance strategy under imperfect inspection.By considering the imperfection of inspections which may introduce uncertainty to components' states and effective ages,a new robust selective maintenance strategy is proposed.The new method consists of three steps: firstly,the posterior probability distributions of components' states and effective ages are inferred by the Bayesian method;secondly,the uncertainty associated with the probability of a system completing a mission is quantified by the expectation and variance of components' reliability measures;thirdly,a multi-objective robust selective maintenance model is formulated and resolved with the aim of maximizing the expectation and minimizing the variance of the probability of a system completing a mission.As demonstrated by a set of numerical examples,the proposed approach can improve the robustness of the mission success effectively.(3)Development of a selective maintenance strategy for systems executing multiple consecutive missions.By extending the traditional selective maintenance strategies for single mission to the case of multiple missions,and the uncertainty associated with components' status and multiple missions is formulated by the corresponding probability distributions.By taking into account of the imperfect maintenance effieicency,a max-min optimization problem is formulated.The simulated annealing-based genetic algorithm is used to solve the resulting non-linear programming problem.Furthermore,the Gaussian quadrature and a discrete numerial method are implemented to improve the computational efficiency of multi-dimensional integration in evaluating probability of a system completing multiple consecutive missions.The results reveal that by the proposed method,the probabilities of a system completing multiple consecutive missions are much greater than the case when the limited maintenance resouces are evenly distributed among multiple missions.