Optimization Of Multi-level Inspection Strategy For Multi-state Systems

Condition monitoring,as a key measure to ensure the safe and reliable operations of complex engineering systems,has been widely implemented to reveal the health status and predict the future states of systems.However,due to the complex working conditions of systems and the limitation of the accuracy of condition monitoring techniques,the inspection data from the condition monitoring oftentimes cannot accurately reveal the true states of systems.On the other hand,the inspection data can be oftentimes collected from multiple physical levels of a system with a hierarchical structure,including the system level,subsystem level and component level.Besides,due to limited inspection resources,such as time,budget,and manpower,it is often impossible to collect inspection data for all components,subsystems,and the entire system simultaneously.Therefore,by taking account of the uncertainty of inspection data,multiple physical levels of systems,and the limited inspection resource,it is a challenge to make the optimal inspection strategy,so as to accurately reveal the health status of complex engineering systems.The key to make the optimal inspection strategy is how to evaluate the effectiveness of a specific inspection strategy on revealing the health status of systems.In this dissertation,we aim at making the optimal multi-level inspection strategy in the context of multi-state systems under the uncertainty of inspection data,multiple physical levels of systems and limited inspection resource.The primary research motivations and innovative points are summarized as follows:(1)Development of an inspection resource optimization model based on multi-level and imperfect inspection data.Firstly,the observation probability matrix is introduced to quantify the uncertainty of inspection data.Then,by taking all possible inspection data as a sample space,an inspection strategy metric is constructed to evaluate the effectiveness of a particular multi-level inspection strategy based on the total probability formula and dynamic reliability assessment.Therefore,an optimization model of multi-level inspection strategy is constructed by maximizing the proposed metric under limited inspection resource.The illustrative examples showed that the optimal inspection strategy found by the proposed method can rationally allocate the inspection resource to improve the effectiveness of revealing the health status of systems.(2)Development of a multi-level inspection strategy optimization model for assuring target mission reliability of multi-state systems.The target mission reliability is defined to determine whether the system can perform the next mission with a desired reliability or not.An inspection strategy metric is,then,proposed to evaluate the confidence of the judgement that the system can meet the target mission reliability under a particular inspection strategy.Therefore,the optimal inspection strategy can be found by maximizing the proposed metric under limited inspection resource.The illustrative examples showed that the proposed method can effectively find the optimal inspection strategy and improve the confidence of judgement that the system can meet the target mission reliability.(3)Development of a multi-level sequential inspection strategy optimization model for multi-state systems.By taking account of all the possible situations of the inspection data with its corresponding true state of systems,an average effectiveness metric of multi-level sequential inspection strategy is proposed for multi-state systems.Then,the value of information of sequential inspection is proposed by comparing the proposed metric with that of without sequential inspections.The illustrative examples showed that the sequential inspection strategy can save inspection resource and improve the effectiveness of inspection.