A neighborhood search relaxation towards railway maintenance optimization

In this paper, we consider the cyclic job scheduling problem of railway vehicles under preventive maintenance to minimize their unavailability, i.e., the total days on maintenance activities of all vehicles in the given time horizon; it is equivalent to the maximization of their total availability. The analysed problem is strongly NP-hard, whereas its solution space is spanned by the combination of permutations. Thus, it causes addition difficulties to traverse it efficiently by local search methods. Therefore, we propose the optimization algorithms, which are based on the Nawaz-Enscore-Ham’s method and the simulated annealing metaheuristic, both equipped with our relaxed search of a neighborhood. The numerical experiments reveal that our idea to enhance the algorithms translates into visible improvements of their efficiency to search solution space and to gain crucial benefits for railway industry in the number of days saved comparing to their earlier basic versions. We also confirm once again the strength and robustness of these applied techniques for various combinatorial optimization problems and possibilities of their further augmentation.