Metaheuristic Optimization of Distributed Job Shop Scheduling: A Study on Encoding Variants and Algorithmic Efficacy

Abstract The Job-Shop Scheduling Problem is defined as the problem of completing the available jobs in a job shop with a limited number of machines in the shortest time possible. This problem is an important and complex optimization problem that businesses, especially in the fields of production and manufacturing, often face and have difficulty in solving in planning for the efficient use of resources. The Distributed Job Shop Scheduling Problem (DJSP), on the other hand, is a more complex problem both in practice and in theory in solving this scheduling problem in multiple facilities. The DJSP is classified as an NP-Hard optimization problem due to its complexity, and it is not operational to solve it with exact solution methods. Due to the size and complexity of the solution space inherent in the problem, it may not be possible to reach an optimal solution. Therefore, an alternative approach, meta-heuristic algorithms, is often used to solve the problem. Meta-heuristic approaches, usually proposed by modeling natural phenomena and intelligent behavior, can provide satisfactory results, although they do not guarantee optimality. In this study, well-known meta-heuristic algorithms are applied to DJSPs with two, three, and four plants. In order to apply the meta-heuristic algorithms to these problems, three encoding schemes proposed in the literature are used. A workload rule approach is used to allocate the jobs in the DJSP according to the number of facilities in an appropriate and efficient manner. The performance of the meta-heuristic algorithms under varying numbers of facilities and encoding schemes is discussed comparatively and supported by statistical evaluations. According to the results, the Artificial Algae Algorithm meta-heuristic algorithm achieved the best results.