A reinforcement learning-guided hyper-heuristic algorithm for cold chain distribution planning

Wu, Baotong, Chen, Zhixiang and Demir, Emrah ORCID: https://orcid.org/0000-0002-4726-2556 2027. A reinforcement learning-guided hyper-heuristic algorithm for cold chain distribution planning. Expert Systems with Applications 333 (Part G) , 134254. 10.1016/j.eswa.2026.134254

PDF - Published Version Available under License Creative Commons Attribution. Download (4MB)

Abstract

Cold chain distribution requires routing decisions that meet service requirements while accounting for operating costs and environmental performance. Because routing choices directly affect travel cost, refrigeration energy use, and carbon emissions, effective planning is essential for sustainable cold chain operations. To address this routing problem, this study proposes a hybrid reinforcement learning-guided hyper-heuristic algorithm (HRL-HHA) that adaptively selects and combines search operators to improve search performance and reduce premature convergence. The proposed dual-layer framework enhances solution quality by coordinating intra- and inter-route search, while the parallel batch Tabu search procedure improves computational efficiency. In addition, we compare six RL-guided HHA variants, including five single-RL variants and a hybrid RL variant. This comparison allows us to examine how different RL-based heuristic selection strategies influence HHA performance. Across the benchmark instances, the proposed method achieves better results than the best solutions reported in the literature. In a real-world case study, it reduces travel distance by approximately 30% and total cost by more than 35%. Further analyses examine fleet selection and carbon pricing sensitivity to provide practical insights for sustainable and efficient cold chain distribution planning.

Item Type:	Article
Date Type:	Publication
Status:	Published
Schools:	Schools > Business (Including Economics)
Publisher:	Elsevier
ISSN:	0957-4174
Date of First Compliant Deposit:	15 September 2026
Last Modified:	15 Sep 2026 10:00
URI:	https://orca.cardiff.ac.uk/id/eprint/189604

Actions (repository staff only)

Edit Item