A Comprehensive Study of Metaheuristic Performance on Job-Shop Scheduling Problems

Le résumé fourni par la source

The article presents an in-depth literature review on the performance of metaheuristics in operations scheduling problems and aims to evaluate the use of metaheuristics, concerning job-shop scheduling problems. In the first part, a literature review was conducted on the significance of operations scheduling and its different types, as well as metaheuristics and jobshop scheduling problems, providing historical context to the three topics. The methodology for the selection of the papers included in the bibliometric study is explained. Twenty articles from Genetic Algorithms, Particle Swarm Optimization, Simulated Annealing and Tabu Search, addressing job-shop problems were selected. Then, various statistical analyses were conducted, such as the analysis of the evolution of results throughout the years and the performance comparison analysis between metaheuristics. Finally, a discussion about the results obtained is held, presenting the conclusions. The statistical analyses revealed that the performance of metaheuristics depends on multiple factors and that their evaluation should not be carried out in isolation. In terms of practical results, the analysis showed that Genetic Algorithms achieved the highest average makespan reduction, followed by Simulated Annealing, Particle Swarm Optimization, and Tabu Search. For example, GA consistently reduced makespan by more than 15% compared to industrial cases, while Tabu Search showed the least consistent performance across studies.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.