Abstract
Effective production scheduling methods are required to realize efficient facility utilization and flexibility for variety production because of diversification of consumer needs and intensification of international competition. This paper proposes an adaptive product release scheduling method using reinforcement learning based a product release scheduler. The scheduler acquires suitable rules that decide product kind and its release timing based on the distribution of the WIP (Work In Process) in the production floor. The proposed method is applied to a multi-product semiconductor manufacturing system and computer simulations are executed to confirm the effectiveness of the proposed method.
