Sustainable Supply Chain Management: A Green Computing Approach Using Deep Q-Networks

Di Yuan,Yue Wan

DOI: https://doi.org/10.1016/j.suscom.2024.101063

2024-11-29

Sustainable Computing: Informatics and Systems

Abstract:This paper addresses the challenges of resource allocation and inventory management in supply chain systems by constructing an intelligent supply chain optimization model based on Deep Q-Networks (ISCO-DQ), emphasizing eco-efficiency. Initially, the study builds a supply chain model that incorporates supplier-customer relationships, guided by the principles of green computing to minimize environmental impact. The model applies Markov Decision Processes to develop a framework for sustainable supplier inventory control, focusing on reducing waste and optimizing resource usage. Utilizing the function approximation capabilities of Deep Q-Networks, the model not only achieves intelligent resource allocation but also prioritizes energy-efficient practices in inventory management. Experimental results indicate that the ISCO-DQ inventory control model converges to approximately -41,400 and -181,300 after around 100 and 300 cycles, respectively, under customer demand distributions that follow normal distributions. Furthermore, compared to traditional single-period stochastic and fixed-order quantity inventory control models, the total cost of the ISCO-DQ model is reduced by an average of 6.7% and 16%, respectively, while minimizing carbon emissions associated with overproduction and excess inventory. Additionally, the ISCO-DQ model significantly mitigates costs arising from demand uncertainty by quickly adapting to fluctuations and optimizing inventory strategies, thereby fostering a circular economy. This demonstrates that the ISCO-DQ inventory control model effectively addresses inefficiencies, inflexibility, and suboptimal resource allocation in conventional supply chain management, ultimately promoting sustainable development and environmental stewardship for enterprises.

computer science, information systems, hardware & architecture