Component-based predictive maintenance modeling for multi-state systems

Component-level predictive maintenance schedules are developed to maximize multi-state system lifetime, considering degrading component multi-state behavior. To maximize the system time-to-replacement, the predictive maintenance schedule is based on individual component performance degradation trends, system performance requirements and component maintenance thresholds. This work is an extension of earlier published research using system-level predictive maintenance. Compared with the system-perspective predictive maintenance model for multi-state system, this model has several advantages. We studied a set of different configuration flow transmission water pipe multi-state systems through both a component and system perspective and compared the results. The conclusion is that the system-perspective model has an advantage when dealing with pure parallel systems; however for other types of system configurations, component-perspective model can provide more efficient maintenance to prolong the estimated lifetime of components and to save maintenance costs. We use a Markov Chain model to obtain the time-dependent state probability for all states of each component, and the universal generating function methodology is used to evaluate the performance of each component and the system. For the examples, a numerical search is used to determine the values of individual component maintenance thresholds to maximize the lifetime of overall system.