A hyper-heuristic for distributed parallel machine scheduling with machine-dependent processing and sequence-dependent setup times

Today, because the market is scattered around the world, manufacturing activities are not limited to a single location and have spread globally. As a result, the discussion of scheduling the factory has changed from a classic single to a network scheduling as a need in the real world. In this regard, this study considers the scheduling of multiple factories by taking into account the job transportation time between factories. The main problem here is that each job would be assigned to which factory and machine. In this research, unrelated parallel machines are considered in which the processing time of jobs depends on the machine and setup time. To minimize the makespan, first, a mixed-integer linear model was proposed in which two types of modeling have been combined. Then, a hyper-heuristic algorithm (HHA) was designed to solve the problem in a reasonable time by choosing the best method among four low-level heuristic methods that are precisely designed according to the properties of the problem. Finally, the efficiency of the proposed algorithm has been compared with the imperialist competitive algorithm (ICA) by conducting experiments. The results show that the proposed algorithm performs very well compared to the ICA and, in more than 75% of the test problems, the proposed algorithm was superior. Also, based on the analysis, in comparing the proposed algorithm with the ICA, it can be concluded that there is a significant difference between the results, and in all cases, the HHA was remarkably better. Considering the challenges and rapid changes of today's market that traditional centralized production planning does not have enough flexibility to respond to them, the results of this research are expected to be useful and attractive for planners in this field.