Optimal failure mode-based preventive maintenance scheduling for a complex mechanical device

This paper addresses the issue of PM scheduling for a complex mechanical with different failure modes. Unlike conventional studies which only consider a single failure mode without prior health information, our work jointly optimizes the failure time estimation and maintenance scheduling for mechanical systems under different failure modes. The proposed approach considers the failure modes which are fatal to the system function and performance. Firstly, the fatal failure modes are identified and divided into two types: the degraded and functional failure modes. The former indicates the one that has an explicit detectable degradation process till the predefined failure threshold is exceeded, while the latter indicates the failure modes that occur suddenly. For both types of failure modes, approaches of failure time estimation are presented to provide the actual health status information for maintenance scheduling. Afterwards, the maintenance scheduling problem is formulated on the estimated failure time distributions of the multiple failure modes. The integer-constrained genetic algorithm (GA) is adopted to optimize the preventive maintenance scheduling by minimizing the life-cycle cost rate. The optimal PM scheduling is derived based on the following: (1) when to perform the intermediate and major maintenance and (2) which failure mode should be maintained at an intermediate maintenance epoch. Finally, the performance of the proposed approach is validated through a real case of the ram subsystem of a boring machine.