Condition-based maintenance optimization for multi-component systems subject to a system reliability requirement

• Published in: Reliability engineering & system safety Vol. 202; p. 107042
• Main Authors: Shi, Yue, Zhu, Weihang, Xiang, Yisha, Feng, Qianmei
• Format: Journal Article
• Language: English
• Published: Barking Elsevier Ltd 01.10.2020; Elsevier BV
• Subjects: Algorithms; Bayesian analysis; Complex systems; Component reliability; Condition-based maintenance; Decision making; Degradation; Dynamic environment; Environmental information; Heuristic methods; Inspection; Multi-component systems; Optimization; Predictive maintenance; Predictive reliability; Preventive maintenance; Reliability aspects; Reliability engineering; System reliability; Predictive reliability; Multi-component systems; Condition-based maintenance; Dynamic environment
• ISSN: 0951-8320; 1879-0836
• DOI: 10.1016/j.ress.2020.107042

•A bayesian updating approach is used to obtain accurate reliability prediction.•A unified maintenance decision framework is proposed.•A two-stage heuristic algorithm is designed to seek optimal maintenance grouping. Prognostic methods for remaining useful life and reliability prediction have been extensively studied in the past decade. However, the use of prognostic information and methods in maintenance decision-making for complex systems is still an underexplored area. In this paper, using a rolling-horizon approach, we develop a condition-based maintenance decision-framework for a multi-component system subject to a system reliability requirement. The system is inspected periodically and new degradation information on components is obtained upon inspection. The new degradation observations are used to update the posterior distributions of the failure model parameters via Bayesian updating, providing more accurate and customized predictive reliabilities. If the predictive system reliability is below the reliability requirement, a novel dynamic-priority-based heuristic algorithm is used to identify a group of components for preventive maintenance. Numerical results show that significant cost savings and improved system reliabilities can be obtained by using more accurate predictive information in maintenance decision-making. We further illustrate the modeling flexibility of the proposed framework by considering dynamic environmental information in decision-making and investigate the potential benefits of incorporating dynamic contexts.