Integrating Artificial Intelligence (AI) with Discrete-Event Simulation (DES) presents an approach for solving complex real-world planning problems. Traditional methods like Integer Linear Programming face limitations when dealing with stochastic production environments characterized by variable process times, equipment breakdowns, and workforce dynamics. The combination of AI and DES overcomes these challenges by enabling detailed modeling of stochastic events while efficiently exploring large solution spaces. Despite successful implementations in job shop scheduling and production planning, existing literature lacks a comprehensive methodology for systematically integrating these technologies. Some authors have already successfully combined Reinforcement Learning (RL) methods with DES for scheduling problems, but do not demonstrate their development using a framework. A novel methodological framework is developed to bridge this gap, providing support for the systematic implementation steps for combining DES with RL. For illustration, the framework is applied to a job rotation problem in a paced assembly line environment.
Kranz et al. (Thu,) studied this question.