Improved Genetic Algorithm Using Reinforcement Learning to Solve the Re-entrant Flexible Flow Shop Scheduling Problem

The Reentrant Flexible Flow Shop Scheduling Problem (RFFSP) involves multiple repetitions of job processing in the production system, which leads to higher scheduling complexity. To solve the RFFSP, this paper proposes a Reinforcement Learning- enhanced Genetic Algorithm (RL-GA). First, a model for the RFFSP is established with the goal of minimizing the total delay time. Then, based on the characteristics of the RFFSP, a machine selection and operation selection (MSOS) encoding scheme is designed. Finally, to address the issue of genetic algorithms getting stuck in local optima and having poor local search capabilities, a reinforcement learning-based method for selecting crossover and mutation strategies is proposed. This allows the algorithm to adaptively choose more suitable crossover and mutation strategies during the iteration process. The proposed method is validated by comparing it with five other scheduling algorithms, demonstrating an improvement in the algorithm's performance