Optimal Maintenance Scheduling of N-Vehicles with Time-Varying Reward Functions and Constrained Maintenance Decisions

Abstract

In this paper we consider the problem of scheduling the maintenance of fleet of N different vehicles over a given planning horizon T. Each vehicle is assumed to have different time-dependant productivity, mean time to repair, and cost of repair with possible limitations on the maximum number of vehicles that can be repaired at a given time epoch. The objective is to maximize the total productivity minus the cost of repairs. We propose an efficient dynamic programming (DP) approach to the solution of this problem. The constraints are translated into feasible binary assignment patterns. The dynamic programming considers each vehicle as a state which takes one of possible feasible patterns. The algorithm seeks to maximize the objective function subject to specific constrains on the sequence of the selected patterns. An example is given and the DP solution is compared with the best of 50,000 randomly selected feasible assignments. The developed algorithm can also be applied to a factory of N production machines, power generators, or car rental.