Metaheuristics for Rolling Stock Cyclic Job Scheduling Problems with Maintenance

In this paper, we analyse rolling stock cyclic job scheduling problems with maintenance to maximize the total availability of vehicles or to minimize the total maintenance cost.The considered issues can be formulated as cyclic job scheduling problems on parallel machines with time and usage based deteriorating effects and maintenance.Since these problems are strongly NP-hard, we propose a dedicated double crossover genetic algorithm and compare its efficiency with other metaheuristics: particle swarm optimization, simulated annealing, and genetic algorithm (using a single crossover).The computational experiments reveal that our approach is robust, optimizes various instances, and overwhelms the other evaluated algorithms for the analysed scenarios correlated with real-life cases.Thus, our new method is applicable for the industrial practice to maximized availability of vehicles as well as to minimize maintenance costs.