Research on Inventory Management Optimization Strategy in Supply Chain Based on Deep Reinforcement Learning

• Published in: 2024 5th International Conference on Computer Vision, Image and Deep Learning (CVIDL) pp. 786 - 791
• Main Authors: Chen, Xinlan, Zheng, Chunting, Liu, Minggao
• Format: Conference Proceeding
• Language: English
• Published: IEEE 19-04-2024
• Subjects: Costs; Deep reinforcement learning; Inventory management; inventory optimization; Stability analysis; supply chain management; Supply chains; Training; Uncertainty
• DOI: 10.1109/CVIDL62147.2024.10603641

At present, the market environment is increasingly complex and changing, and the traditional inventory management strategy is difficult to adapt to the changes in the supply chain. For this reason, firstly, this paper proposes a management optimization framework applicable to supply chain inventory control based on deep reinforcement learning theory. The framework is able to learn the optimal strategy through an end-to-end neural network. Secondly, multiple mechanisms are introduced to enhance the training efficiency and stability, so that it can better adapt to the complexity and variability of the supply chain. Finally, simulation experiments are carried out in multi-species and multi-warehouses, and the results show that the framework can significantly outperform the traditional heuristic and linear planning methods in terms of total cost, out-of-stock rate, inventory turnover, etc., and exhibits superior adaptability and robustness.