Learning Flexible Job Shop Scheduling under Limited Buffers and Material Kitting Constraints
#Flexible Job Shop Scheduling #Limited Buffers #Material Kitting #Deep Reinforcement Learning #Heterogeneous Graph Network #Buffer Utilisation #Pallet Changes #Makespan #Scheduling Heuristics #Production Line Simulation
📌 Key Takeaways
- Authors combine deep reinforcement learning with a heterogeneous graph neural network to model global interactions in a job shop context.
- The study extends the classic flexible job shop scheduling problem by adding realistic constraints: limited buffer capacity and material‑kitting (pallet grouping).
- The model’s message‑passing framework is designed to avoid decisions that trigger frequent pallet changes, thereby improving buffer utilisation and overall schedule quality.
- Experimental results on synthetic and real production‑line data show the proposed method outperforms traditional scheduling heuristics and advanced DRL techniques in terms of makespan and pallet‑change frequency.
- A supplementary video provides a visual simulation of the production line that illustrates the schedule’s evolution and the impact of the algorithm on buffer usage.
📖 Full Retelling
The paper *Learning Flexible Job Shop Scheduling under Limited Buffers and Material Kitting Constraints* was authored by Shishun Zhang, Juzhan Xu, Yidan Fan, Chenyang Zhu, Ruizhen Hu, Yongjun Wang, and Kai Xu, and was submitted to the arXiv preprint server (cs.AI) on 27 February 2026. It tackles the Flexible Job Shop Scheduling Problem (FJSP) – a well‑known model in manufacturing – but explicitly incorporates two practical constraints that are often abstracted: limited on‑hand buffer capacity and material‑kitting (the grouping of parts on pallets). By combining deep reinforcement learning with a heterogeneous graph neural network that captures interactions among machines, operations, and buffers, the authors show that their method reduces makespan and pallet‑changing events more effectively than both traditional heuristic schedulers and existing DRL baselines. Experiments on synthetic data and a real production‑line dataset confirm superior performance, and a supplementary video demonstrates a simulation that visualises the scheduling process in real time.
🏷️ Themes
Artificial Intelligence, Deep Reinforcement Learning, Graph Neural Networks, Manufacturing Scheduling, Production‑Line Optimization
Entity Intersection Graph
No entity connections available yet for this article.
Original Source
--> Computer Science > Artificial Intelligence arXiv:2602.24180 [Submitted on 27 Feb 2026] Title: Learning Flexible Job Shop Scheduling under Limited Buffers and Material Kitting Constraints Authors: Shishun Zhang , Juzhan Xu , Yidan Fan , Chenyang Zhu , Ruizhen Hu , Yongjun Wang , Kai Xu View a PDF of the paper titled Learning Flexible Job Shop Scheduling under Limited Buffers and Material Kitting Constraints, by Shishun Zhang and 6 other authors View PDF HTML Abstract: The Flexible Job Shop Scheduling Problem originates from real production lines, while some practical constraints are often ignored or idealized in current FJSP studies, among which the limited buffer problem has a particular impact on production efficiency. To this end, we study an extended problem that is closer to practical scenarios--the Flexible Job Shop Scheduling Problem with Limited Buffers and Material Kitting. In recent years, deep reinforcement learning has demonstrated considerable potential in scheduling tasks. However, its capacity for state modeling remains limited when handling complex dependencies and long-term constraints. To address this, we leverage a heterogeneous graph network within the DRL framework to model the global state. By constructing efficient message passing among machines, operations, and buffers, the network focuses on avoiding decisions that may cause frequent pallet changes during long-sequence scheduling, thereby helping improve buffer utilization and overall decision quality. Experimental results on both synthetic and real production line datasets show that the proposed method outperforms traditional heuristics and advanced DRL methods in terms of makespan and pallet changes, and also achieves a good balance between solution quality and computational cost. Furthermore, a supplementary video is provided to showcase a simulation system that effectively visualizes the progression of the production line. Comments: 8 pages, 8 figures, conference Subjects: Artificial ...
Read full article at source