- Titel
- A Comparative Simulation Study of Deep Reinforcement Learning and Dispatching Rules for a Complex Job Shop
- Zitierfähige Url:
- https://nbn-resolving.org/urn:nbn:de:bsz:14-qucosa2-996259
- Übersetzter Titel (DE)
- Eine vergleichende Simulationsstudie zu Deep-Reinforcement-Learning und Dispatching-Regeln für einen komplexen Job-Shop
- Konferenz
- 21. ASIM-Fachtagung Simulation in Produktion und Logistik. Dresden, Germany, 24. bis 26. September 2025
- Quellenangabe
- Simulation in Produktion und Logistik 2025
Herausgeber: Sebastian Rank
Herausgeber: Mathias Kühn
Herausgeber: Thorsten Schmidt
Erscheinungsort: Dresden
Verlag: Technische Universität Dresden, Professur für Technische Logistik
Erscheinungsjahr: 2025
ISBN: 978-3-86780-806-4 - Erstveröffentlichung
- 2025
- DOI
- https://doi.org/10.25368/2025.237
- Abstract (EN)
- In complex job shop environments such as semiconductor wafer fabrication facilities (wafer fabs), dispatching rules (DRs) play a critical role in managing shop floor operations. While individual DRs offer specific strengths, they also exhibit inherent limitations. To combine their advantages, hybrid dispatching strategies that integrate multiple single rules are commonly employed. Traditionally, such hybrid rules rely on static weight assignments to determine the contribution of each rule. However, static weighting schemes lack adaptability to the dynamic nature of shop floor conditions. With the advancement of machine learning, particularly deep reinforcement learning (DRL), it has emerged as a promising approach for dynamic and adaptive decision-making. In this study, we propose a novel DRL approach based on the Actor-Critic framework that dynamically assigns weights to individual DRs, thereby constructing a context-aware hybrid rule. To evaluate the effectiveness of the proposed approach, simulation experiments are conducted using a representative wafer fabs model. Simulation plays a crucial role by providing a safe, flexible, and controlled environment to test dispatching strategies under realistic production variability. The simulation results demonstrate that the proposed DRL approach consistently outperforms both traditional DRs and hybrid rules with static weight assignments in terms of reducing cycle time, improving on-time delivery, and maintaining high throughput. These findings highlight the potential of DRL, combined with simulation-based evaluation, to enable intelligent, adaptive dispatching in complex manufacturing systems.
- Freie Schlagwörter (DE)
- Simulation, Künstliche Intelligenz, Modellierung, Optimierung, Simulationsstudie, Dispatching-Regel
- Freie Schlagwörter (EN)
- Simulation, Artificial Intelligence, Modelling, Optimisation, Simulation study, dispatching rule
- Klassifikation (DDC)
- 620
- Klassifikation (RVK)
- ZM 9500
- Publizierende Institution
- Technische Universität Dresden, Dresden
- Version / Begutachtungsstatus
- publizierte Version / Verlagsversion
- URN Qucosa
- urn:nbn:de:bsz:14-qucosa2-996259
- Veröffentlichungsdatum Qucosa
- 18.10.2025
- Dokumenttyp
- Konferenzbeitrag
- Sprache des Dokumentes
- Englisch
- Lizenz / Rechtehinweis
CC BY 4.0