A Q-Learning-Based Hyper-heuristic Algorithm for the Muti-Objective Integrated Scheduling of Distributed Production and Delivery Problem

The integrated scheduling problem has gained widespread application in the manufacturing and logistics industries due to its notable advantages in resource integration, cost control, and enhanced response speed. A Q-learning-based hyper-heuristic algorithm (QLHH) is proposed to solve the integrated scheduling of distributed production and delivery problem (ISDPDP) in this study. By deeply integrating Q-learning with the hyper-heuristic algorithm, the QLHH algorithm rapidly selects low-level heuristics using a Q-table and an improved greedy algorithm to explore the solution space. Energy-saving is fully considered in the production stage. The states and actions based on problem characteristics are designed to balance energy consumption with multi-objective scheduling and guide the global search. The QLHH accurately captures the core of the problem and addresses the multi-objective integrated scheduling challenge by designing low-level heuristics and energy-saving strategies based on problem knowledge. Experiments demonstrate that the QLHH algorithm outperforms comparison algorithms on test datasets, providing a practical solution for multi-objective integrated scheduling in distributed production and delivery problem.