University of Twente Student Theses
Exploring the potential of deep Q learning on solving the dynamic storage location assignment problem
Vries, J. D. de (2024) Exploring the potential of deep Q learning on solving the dynamic storage location assignment problem.
|
PDF
1MB |
| Abstract: | This research tackles the dynamic storage location assignment problem with a specialised deep reinforcement learning algorithm. The formulation of the dynamic storage location assignment problem and the deep reinforcement learning aim to build a bridge between theory and practice. The objective of this research is to decrease the order picking time in a warehouse by continuously optimising the storage location assignment. Periodic ABC classification is applied to create benchmark solutions, which are used as performance comparisons. The dynamic storage location assignment problem is a complex problem, that requires an intelligent algorithm to be solved. In this research, several steps are taken to present a solution to the problem at hand. First, the dynamic storage location assignment problem is formulated as a mixed-integer linear program. Then, the formulated program is translated into a Markov decision process. On top of that, a deep Q learning algorithm is formulated and programmed to tackle the formulated Markov decision process. Several experiments are created to obtain a complete perspective on the performance of the deep Q learning algorithm. Finally, the results are presented and analysed, showing that the formulated deep Q learning algorithm does not outperform the proposed benchmarks. However, the thorough analysis of the results provide concrete improvements and other future research directions, that have the potential to improve the performance in the future. |
| Item Type: | Essay (Master) |
| Clients: | Supply Value, Zeist, Netherlands |
| Faculty: | BMS: Behavioural, Management and Social Sciences |
| Subject: | 54 computer science |
| Programme: | Industrial Engineering and Management MSc (60029) |
| Link to this item: | https://purl.utwente.nl/essays/103460 |
| 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