• 목록

In multistage manufacturing processes, achieving high-quality outputs while minimizing product cycle time is a critical challenge. This study presents an approach to scheduling in multistage manufacturing by using reinforcement learning. The proposed method integrates an environment module, which models machine status and queue dynamics, with a product agent module that governs the actions of individual product agents. To enhance productivity, product agents dynamically interact with the environment module, making optimized decisions to maintain stable cycle time for each product. Simultaneously, a quality prediction model, developed using machine learning, evaluates the final rewards, ensuring quality considerations are embedded in the scheduling process. By adopting this dual-focus strategy, the framework effectively balances productivity and quality, optimizing path selection for multistage manufacturing environments. This approach provides a solution for efficient and adaptive manufacturing scheduling.