A Q-Learning Based Hyper-Heuristic Algorithm for the Energyefficient Distributed Assembly Fuzzy Flexible Job-Shop Scheduling Problem

In this article, for the Energy-efficient distributed assembly fuzzy flexible job-shop scheduling problem (EE-DAFFJSP), we establish a problem model to minimize the maximum completion time and the total energy consumption and propose a hyperheuristic Q-learning based hyper heuristic algorithm (QLHHA) to solve the problem. Firstly, a three-dimensional vector coding rule based on process sequences are designed, two initialization strategies are used to improve the quality of the initial solution. Then, in order to improve the guidance of the search direction, Q-learning algorithm is used to learn and accumulate the information of highquality alignments, which consist of 8 low-level heuristics (LLH) (i.e., 10 effective neighborhood operations) designed by combining the characteristics of the problem; and simulated annealing-based mechanism is added in the process of perturbation as a new heuristic method to perform the search. Finally, QLHHA is verified to be effective in solving EE-DAFFJSP through simulation experiments and algorithm comparison.