An Adaptive Hyper‐Heuristic Framework for the Hub Location Problem in Intelligent Supply Chain Networks

ABSTRACT

The hub location problem (HLP) is a central challenge in strategic supply chain design, requiring the optimal placement of hub facilities and the routing of goods to minimise overall transportation costs. Due to its combinatorial complexity and dynamic constraints, such as fluctuating demand, transportation costs and capacity limits, traditional solution techniques often fall short in providing scalable and adaptive performance. This paper proposes a novel reinforcement learning–based hyper‐heuristic framework designed to address both capacitated and uncapacitated HLP variants under dynamic supply chain scenarios. The framework operates by intelligently selecting among a pool of low‐level heuristics, guided by a high‐level learning strategy that adapts based on feedback from the search space. Extensive experiments using benchmark datasets and simulated real‐world data demonstrate that the proposed method outperforms traditional heuristics and metaheuristics in terms of solution quality, computational efficiency and adaptability. The results underscore the potential of hyper‐heuristics as a generalisable and scalable approach for solving complex logistics optimisation problems in evolving operational environments.