國 立 陽 明 交 通 大 學 演 講

講    題：半導體製造中混合產品生產模式的強化學習運行控制算法

(Reinforcement Learning Run-to-Run Control Algorithm for Mixed-Product Production Mode in Semiconductor Manufacturing)

主 講 人：范書愷(Shu-Kai Fan)

(國立臺北科技大學 工業工程與管理系)

主 持 人：董弘平教授

主辦單位：陽明交通大學工業工程與管理系

時間：113年11月25日(星期一) 13:20~15:10   

地    點：管理二館MB520教室(5F)

演講摘要：

This research introduces an innovative Run-to-Run (R2R) control framework leveraging the Deep Deterministic Policy Gradient (DDPG) algorithm, tailored for mixed-product production in semiconductor manufacturing. The proposed framework adapts the DDPG algorithm to create a deep reinforcement learning environment optimized for the unique challenges of mixed-product production modes. To enhance

the effectiveness of the DDPG model, three advanced mechanisms are integrated: a piecewise reward function, dynamic target training, and the novel recall principle. Comprehensive simulation results demonstrate that the proposed R2R control framework significantly outperforms five well-established mixed-product R2R control algorithms. These findings highlight the potential of deep reinforcement learning to handle complex, dynamic environments with continuous action spaces, offering a robust solution for mixed-product R2R control in semiconductor manufacturing.

演講性質：學術研究專題

歡迎聽講