Genetic algorithm and simulation based hybrid approach to production scheduling

The production scheduling problem is a practical job-shop problem with processing constraints that are more restrictive and a scheduling objective. As one of the major constraints in genetic algorithm (GA) models, operation time has a deterministic solution. However, in real systems, due to various kinds of uncertain factors such as queuing, breakdowns and repairing time of machines, the optimal solution in the GA procedure cannot correctly represent the stochastic behaviour of a real operation. To solve this problem, a hybrid approach involving the GA and a simulation is presented. In this study, the GA is used for optimization of schedules, and the simulation is used to minimize the maximum completion time for the last job with fixed schedules from the GA model. We obtain more realistic production schedules with an optimal completion time reflecting stochastic characteristics by performing the iterative hybrid GA - simulation procedure. It has been shown that the hybrid approach is powerful for complex production scheduling.