University of Twente Student Theses
Deep Reinforcement Learning for the Stochastic Joint Replenishment Problem with Non-Zero Lead Time
Nicklin, Jonathan (2025) Deep Reinforcement Learning for the Stochastic Joint Replenishment Problem with Non-Zero Lead Time.
|
PDF
2MB |
| Abstract: | Joint replenishment is a complex problem within inventory management, focusing on defining ordering policies in a multi-item setting. Historically it has been approached by generating rule-based policies to determine how much of each item should be included after an order is triggered, but recent effort has been placed in using advances in machine learning within the field. This work contributes to the existing body of knowledge by developing a generalised neural network joint replenishment policy by incorporating demand stochasticity and non-zero lead time durations in ordering at most one unit of transport per period. Applying deep controlled learning to a specialised setting pertaining to dominant transport costs, the proposed solution vastly outperforms existing approaches to learn a policy which effectively exploits cost parameters and maintains a high fill rate for a small item setting. Expanding to a larger problem instance, its computational efficiency is evidenced to impair its ability to perform well. The presented research exhibits potential for further effort to be placed in neural network solutions for general joint replenishment applications. |
| Item Type: | Essay (Master) |
| Faculty: | BMS: Behavioural, Management and Social Sciences |
| Subject: | 31 mathematics, 50 technical science in general, 54 computer science |
| Programme: | Industrial Engineering and Management MSc (60029) |
| Link to this item: | https://purl.utwente.nl/essays/106427 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page