Simulation optimization for the MRO scheduling problem based on multi-fidelity models

Maintenance, repair and overhaul (MRO) are important segments of the remanufacturing industry for the maintenance of complex capital goods. The uncertain processing time and routings make MRO scheduling differ greatly from traditional manufacturing. Simulation optimization is suitable to solve the problems with uncertainty. This work uses multi-fidelity optimization with ordinal transformation and optimal sampling framework to solve a MRO scheduling problem. We build a high-fidelity simulation model which is stochastic and time-consuming. Instead of directly running the high-fidelity model, we provide a deterministic low-fidelity model to transform the original solution space into a one-dimensional ordinal space. The transformed solution space is partitioned into several groups. An efficient optimal computing budget allocation method is used to sample within these groups. Numerical results and comparisons show that this framework is computationally effective to handle those uncertainties, and provides high quality schedules with the low tardiness for the MRO scheduling problem.