International Journal of Mechanical Engineering

With the rapid development of Industry 4.0 and intelligent manufacturing, the optimization of mechanical processing workshop scheduling has become a key link to enhance production efficiency and reduce energy consumption. This paper addresses the insufficiency of traditional workshop scheduling methods in dynamic environments and proposes an intelligent scheduling framework that integrates multi-objective optimization and reinforcement learning. Firstly, based on a systematic analysis of domestic and international literature on intelligent manufacturing scheduling (including Just-In-Time (JIT), Flexible Job Shop Scheduling Problem (FJSP), digital twin technology, etc.), a mixed-integer linear programming (MILP) model is constructed with the objectives of minimizing the maximum completion time (Makespan), balancing equipment load, and reducing energy consumption. Secondly, to address the NP-hard nature of the model, an improved NSGA-III multi-objective genetic algorithm is designed, introducing an adaptive crossover operator and a Pareto front screening mechanism based on machine learning to enhance convergence efficiency.

Intelligent manufacturing; Workshop scheduling optimization; Multi-objective optimization; Deep reinforcement learning; Energy consumption optimization