A cooperative Q-learning-based memetic algorithm for distributed assembly heterogeneous flexible flowshop scheduling

• Published in: Expert systems with applications Vol. 288; p. 128198
• Main Authors: Deng, Jiawen, Zhang, Jihui, Yang, Shengxiang
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.09.2025
• Subjects: Assembly; Distribution production; Memetic algorithm; Q-learning; Distribution production; Memetic algorithm; Q-learning; Assembly
• ISSN: 0957-4174
• DOI: 10.1016/j.eswa.2025.128198

With the continuous development of advanced technologies, turbulent market environments and heterogeneity of customer requirements make manufacturing more complicated; therefore efficient organization of distributed production and assembly is indispensable. However, seldom research considers setup time, transportation, and learning effect simultaneously, while they frequently influence assembly efficiency. In this paper, a distributed assembly heterogeneous flexible flowshop scheduling problem with sequence-dependent setup time, transportation, and learning effect (DAHFFS-STL) is investigated. Meanwhile, timely completion is imperative for a firm corporate reputation. Afterward, a cooperative Q-learning-based memetic algorithm (CQLMA) is devised to tackle this problem to optimize the minimum total weighted earliness and tardiness (TWET). In CQLMA, first, a group of constructive heuristics is utilized to initialize the population. Second, multiple crossover and mutation operations are implemented to expand the search space and improve the global search ability. Third, tabu search is utilized to further strengthen the exploitation ability. Subsequently, the Q-learning algorithm is leveraged to dynamically select suitable operators, thereby enhancing the optimization ability of the CQLMA. Finally, exhaustively comparisons affirm that the CQLMA has achieved more significant performance in handling DAHFFS-STL.