A Hyper-Heuristic Algorithm with Q-Learning for Distributed Permutation Flowshop Scheduling Problem

Abstract

The Distributed Permutation FlowShop Scheduling Problem (DPFSP) is a challenging combinatorial optimization problem with many real-world applications. This paper proposes a Hyper-Heuristic Algorithm with Q-Learning (HHQL) approach to solve the DPFSP, which combines the benefits of both Q-learning and hyper-heuristic techniques. First, based on the characteristics of DPFSP, a DPFSP model is established, and coding scheme is designed. Second, six simple but effective low-level heuristics are designed based on swapping and inserting jobs in the manufacturing process. These low-level heuristics can effectively explore the search space and improve the quality of the solution. Third, a high-level strategy based on Q-learning was developed to automatically learn the execution order of low-level neighborhood structures. Simulation results demonstrate that the proposed HHQL algorithm outperforms existing state-of-the-art algorithms in terms of both solution quality and computational efficiency. This research provides a valuable contribution to the field of DPFSP and demonstrates the potential of using Hyper-Heuristic techniques to solve complex problems.