Dynamic Reliability Assessment For Multi-state Systems By Aggregating Multi-level Inspection Data

Multi-state is one of the characteristics of complex engineered systems. The multistate reliability theory has become an emerging research topic in recent years as it is able to characterize the complicated deteriorating process of systems in a finer fashion. Nevertheless, it is noteworthy that most existing studies of multi-state system reliability theory are on the basis of traditional time-based reliability models with which reliability measures of an MSS are assessed from a “population” or “statistic” perspective. With such treatment, uncertainties associated with the heterogeneous deterioration behaviors from one piece of system to another are all encompassed by the “population” information. These methods, however, fail to characterize the stochastic behaviors of a specific individual multi-state system. For a specific individual system, if additional useful knowledge related to the stochastic deteriorating behaviors of the system becomes available, the uncertainty associated with the heterogeneity can be further reduced, so as to improve the accuracy of reliability assessment for this specific system. This is called dynamic reliability assessment.As dynamic reliability assessment is capable of assessing health status, tracking degradation trend, and predicting future reliability for an individual system, it has become an emerging and important research topic of reliability engineering in recent years. Undoubtedly, aggregating the inspection data from multiple levels of a system may bring benefit to dynamic reliability assessment method. This thesis devotes to aggregate inspection data from multiple levels of a system to achieve a better assessment of dynamic reliability for a specific individual multi-state system. Additionally, an optimal selective maintenance strategy of an MSS is put forth on the basis of dynamic reliability assessment. The major contributions of this thesis are summarized as follows:(1) Development of a dynamic reliability assessment method for MSSs based on imperfect system-level inspection data. With the consideration of the heterogeneity of deteriorating processes of multi-state systems due to diverse operating loads and environments experienced by each individual system, this thesis developed a dynamic reliability assessment method based on imperfect system-level inspection data. The numerical example shows that the proposed approach yields more accurate reliability prediction than the traditional time-based reliability models.(2) Development of a dynamic reliability assessment method by aggregating inspection data across multiple levels of an MSS. The inspection data of a system can be collected from multiple levels, say system-level, subsystems-level, and component-level, reflecting the health condition of the system from different physical levels. A dynamic reliability assessment model by aggregating multi-level inspection data is introduced. To take into account of the imperfection of inspection data, a set of two-stage recursive Bayesian formulations is developed to dynamically update the reliability function of a specific MSS over time. The proposed method is demonstrated by two numerical studies.(3) Investigation of a selective maintenance strategy for MSSs based upon the dynamic reliability assessment method. A novel imperfect maintenance model for multistate components is proposed. Based upon the dynamic reliability assessment method developed earlier in our work, a selective maintenance optimization problem is investigated to examine how the selective maintenance strategy can be improve by taking account of system-level inspection data. The genetic algorithm is employed to solve the optimization problem. Numerical studies show that the proposed method produces better results than the case where system/component states are not inspected.