Opportunistic Replacement Optimization for Multi-Component System Based on Programming Theory

It is widely accepted that too excessive or too insufficient maintenance actions on a system are consumptive or potentially risky. This paper focuses on the optimization of opportunistic replacement for a multicomponent system in which no failure or suspension histories can be used for prediction of all the critical components in the system. Firstly, the remaining useful life (RUL) is predicted using the real-time sensor data, which is based on an “individual-based lifetime inference” method. Then a failure risk estimation method is introduced, which is based on the degradation extent and service time of components. Subsequently, the possible replacement combinations of components are compared, which is based on a proposed current-term cost rate. Finally, the best replacement scheduling is selected. The proposed framework is validated by the simulation dataset and PHM-2012 competition bearing dataset. Group replacement and individual replacement are conducted for comparison, and sensitivity analysis is discussed.