An Improved Differential Evolution Algorithm Based on Q-Learning for Solving Flexible Assembly Job Shop Scheduling Problem

To address the flexible assembly job shop scheduling problem, an improved differential evolution algorithm based on Q-Learning is proposed to optimize the makespan, the average product completion time and the total machine load. Firstly, a multi-objective mixed integer programming model is constructed by considering the constraints of workpiece processing sequence and assembly matching. Secondly, a hybrid population initialization strategy is designed to balance the diversity and quality of the initial solution by combining random generation with priority rules. On this basis, the Q-learning mechanism is introduced to adaptively adjust the crossover probability and mutation probability of the differential evolution algorithm, and the search ability of process coding and machine coding is dynamically optimized, so as to improve the balance between global exploration and local development of the algorithm.