Joint optimization of multi-stage component reassignment and preventive maintenance for balanced systems considering imperfect maintenance

Abstract

In this paper, a balanced system consisting of several multi-state components is considered to work during multiple sequential stages. The system is balanced when the largest state difference among all components is no greater than a predetermined value. The system fails when it loses balance or when at least one component is completely failed. Components on different positions are functionally exchangeable, while their degradation characteristics vary with distinct loads and external environments. To keep system balance in the long term and improve its reliability, a joint policy of component reassignment and preventive maintenance is proposed considering imperfect maintenance. The positions of components can be exchanged, and they can be maintained to any better states or be left in the present states. At the beginning of each stage, operators need to decide whether and how to reassign and maintain the components to minimize the total expected cost. A Markov decision process is constructed to model the system operation process and to obtain the optimal joint strategies. Two heuristic policies are presented to compare with the proposed policy. A numerical example based on a three-wheel mobile robot is investigated to illustrate the proposed model and validate the effectiveness of the algorithm.