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

This chapter explores the application of a Q-learning-based hyper-heuristic algorithm (QLHH) to tackle the integrated scheduling of distributed production and delivery problems (ISDPDP). The text delves into the current challenges and complexities of distributed production and delivery models, highlighting the need for effective scheduling strategies to maximize overall benefits. The proposed QLHH algorithm leverages Q-learning, an improved epsilon-greedy action selection strategy, and low-level heuristics tailored to the characteristics of ISDPDP. The chapter provides a comprehensive overview of the algorithm's framework, including the agent, environment, state, action, and reward mechanisms. It also discusses the design of low-level heuristics and their role in exploring the solution space. Experimental results demonstrate the algorithm's performance across various test instances, with comparisons to other algorithms such as NS, SS, and RIG. The results indicate that the QLHH algorithm exhibits good robustness and stable performance, making it a promising solution for large-scale integrated scheduling problems. The chapter concludes with a discussion on future research directions, including further optimization of the QLHH mechanism and the development of multiple-objective optimization models.