TY - GEN
T1 - On Scheduling a Photolithograhy Toolset Based on a Deep Reinforcement Learning Approach with Action Filter
AU - Kim, Taehyung
AU - Kim, Hyeongook
AU - Lee, Tae Eog
AU - Morrison, James Robert
AU - Kim, Eungjin
N1 - Funding Information:
This work was supported by Samsung Display Co., Ltd.
Publisher Copyright:
© 2021 IEEE.
PY - 2021
Y1 - 2021
N2 - Production scheduling of semiconductor manufacturing tools is a challenging problem due to the complexity of the equipment and systems in modern wafer fabs. In our study, we focus on the photolithography toolset and consider it as a non-identical parallel machine scheduling problem with random lot arrivals and auxiliary resource constraints. The proposed methodology strives to learn a near optimal scheduling policy by incorporating WIP, masks, and the tardiness of jobs. An Action Filter (AF) is proposed as a methodology to eliminate illogical actions and speed the learning process of agents. The proposed model was evaluated in a simulation environment inspired by practical photolithography scheduling problems across various settings with reticle and qualification constraints. Our experiments demonstrated improved performance compared to typical rule-based strategies. Relative to our learning methods, weighted shortest processing time (WSPT) and apparent tardiness cost with setups (ATCS) rules perform 28% and 32% worse for weighted tardiness, respectively.
AB - Production scheduling of semiconductor manufacturing tools is a challenging problem due to the complexity of the equipment and systems in modern wafer fabs. In our study, we focus on the photolithography toolset and consider it as a non-identical parallel machine scheduling problem with random lot arrivals and auxiliary resource constraints. The proposed methodology strives to learn a near optimal scheduling policy by incorporating WIP, masks, and the tardiness of jobs. An Action Filter (AF) is proposed as a methodology to eliminate illogical actions and speed the learning process of agents. The proposed model was evaluated in a simulation environment inspired by practical photolithography scheduling problems across various settings with reticle and qualification constraints. Our experiments demonstrated improved performance compared to typical rule-based strategies. Relative to our learning methods, weighted shortest processing time (WSPT) and apparent tardiness cost with setups (ATCS) rules perform 28% and 32% worse for weighted tardiness, respectively.
UR - http://www.scopus.com/inward/record.url?scp=85126095083&partnerID=8YFLogxK
U2 - 10.1109/WSC52266.2021.9715450
DO - 10.1109/WSC52266.2021.9715450
M3 - Conference contribution
AN - SCOPUS:85126095083
T3 - Proceedings - Winter Simulation Conference
BT - 2021 Winter Simulation Conference, WSC 2021
PB - Institute of Electrical and Electronics Engineers Inc.
Y2 - 12 December 2021 through 15 December 2021
ER -