A Novel Multi-Objective Memetic Algorithm for Multi-Task Fuzzy Flexible Job Shop Scheduling

Multi-task multi-objective fuzzy flexible job shop scheduling (FFJSS), different from traditional multi-objective FFJSS, offers the advantage of simultaneously addressing distinct multi-objective scheduling tasks. Evolutionary multi-task is a popular optimization paradigm conserves computational resources and enhances algorithm performance by sharing and leveraging knowledge between different tasks. However, we face the challenges of both multi-task and multi-objective optimization. To address these challenges, we propose a novel multi-objective memetic algorithm (MOMA). First, we design a unified representation with two vectors for both routing and sequencing decisions, which leads to a unified search space within the multifactorial framework. We then propose a dual-neighborhood multifactorial transfer strategy by adjusting the size of the neighborhoods adaptively, which can effectively balance the convergence and diversity of the population and identify transferrable solutions. Additionally, an improved neighborhood update strategy is proposed to improve the rate of positive transfer. To verify the effectiveness of the proposed algorithm, eight distinct multi-task FFJSS scenarios are constructed based on existing FFJSS datasets. The results show that MOMA consistently outperformed the state-of-the-art algorithms across a wide range of test instances. Further analyses of the transfer process have shown that complex tasks can benefit from simple tasks. .