Research on the Application of Evolutionary Algorithms Enhanced by Deep Learning in Intelligent Manufacturing Scheduling

Intelligent scheduling holds a significant position in modern manufacturing, becoming a challenging problem due to its complexity and multiple constraints. This study proposes a novel reinforcement learning approach that combines Actor-Critic Algorithm and Genetic Algorithm, modeling scheduling problems as multi-agent Markov decision processes and comprehensively considering factors such as completion time, machine workload, and computational efficiency. The method utilizes a parallel strategy to enhance global search capabilities, thus improving computational efficiency. Through Actor-Critic Algorithm, it adaptively adjusts crossover and mutation rates, thereby reducing decision-making time and lowering labor costs. Experimental results demonstrate significant improvements in decision quality and efficiency compared to traditional rules and metaheuristic algorithms, providing new insights and theoretical support for solving large-scale flexible job shop scheduling problems.