Multiple-Trial/Best-Move Simulated Annealing for Flexible Job Shop Scheduling with Scarce Setup-Operators

Abstract

In this paper, we developed a multi-trial/best-move simulated annealing for a dual resource-constrained flexible job shop scheduling problem (DRC-FJSP). The algorithm creates many neighborhood solutions at each iteration by perturbing the leading solution. Then, the best solution generated is compared with the current leading solution for probabilistic acceptance or rejection. The algorithm outperforms the classical simulated annealing in terms of convergence speed while solving the considered DRC-FJSP. Unlike many papers in DRC-FJSP, which assumed machine operators as scarce resources, we consider skilled setup operators as a constraining resource instead. This is because machine operators are becoming less constraining due to increased adoption of automation and numerical controlled machines. As a result, machine operators become machine tenders who do not require the skills to perform the step-by-step production process. Thus, the assumption of those low skill labors as constraining resources can no longer be justified. Skilled setup operators, on the other hand, continue to become constraining resources as their skill cannot be easily automated, particularly with the increased complexity of automation in manufacturing.