Daily Arxiv

This is a page that curates AI-related papers published worldwide.
All content here is summarized using Google Gemini and operated on a non-profit basis.
Copyright for each paper belongs to the authors and their institutions; please make sure to credit the source when sharing.

Constraint Programming Models For Serial Batch Scheduling With Minimum Batch Size

Created by
  • Haebom

Author

Jorge A. Huertas, Pascal Van Hentenryck

Outline

This paper addresses the serial batch (s-batch) scheduling problem in a parallel machine environment. In particular, we consider the case where task weights and start times are different and there is a sequence-dependent setup time between different task families. Previous studies have not considered the minimum batch size constraint, but this paper solves the s-batching problem including the minimum batch size constraint, which is commonly found in real environments such as semiconductor manufacturing and metal industries, by using the constraint programming (CP) technique. We propose three CP models (interval assignment model, global model, and hybrid model) and evaluate their performance by comparing them with the existing mixed integer programming (MIP) model.

Takeaways, Limitations

•
Takeaways:
◦
We present a novel method to effectively solve the s-batching problem with minimum batch size constraints using constraint programming (CP).
◦
The proposed three CP models are applicable to various s-batching variant problems, and are experimentally proven to find better solutions faster than the existing MIP models.
◦
Contributes to solving scheduling problems considering minimum batch size constraints in real industrial environments (semiconductor manufacturing, metal industry, etc.).
•
Limitations:
◦
The performance of the proposed CP model may vary depending on the specific problem instance.
◦
Further research is needed on the scalability of CP models to more complex constraints or larger problems.
◦
Comparative analysis with the MIP model used in this paper may be limited to a specific MIP model. Additional comparative analysis with other MIP models may be required.
👍
Made with Slashpage