Flexible Job Shop Scheduling Based on Energy Consumption of Method Research

In current manufacturing processes, the importance of sustainable development concepts in helping enterprises reduce costs and improve efficiency is increasing. The workshop scheduling problem considering energy consumption has become a meaningful research direction. This paper focuses on the solution methods for the flexible job-shop scheduling problem (FJSP) considering energy consumption. Deep reinforcement learning, with its self-learning and adaptive characteristics, has become an important means for solving scheduling problems. By establishing a multi-objective optimization model aimed at minimizing the maximum completion time and energy consumption, this paper solves the flexible job-shop scheduling problem considering energy consumption (GFJSP) based on an improved deep reinforcement learning algorithm, D3QN. This paper proposes an improved MD3QN algorithm by introducing a dual experience pool, a noisy neural network, and a reward mechanism combining single-step and multi-step rewards on the basis of the D3QN algorithm, thereby further improving the algorithm’s convergence, generalization ability, and solution stability. The effectiveness of the algorithm improvement is verified through a test dataset expanded based on standard instances, and the practical application effect is validated through a case study using the production data of hub bearings from a bearing company. The improved effectiveness of the MD3QN algorithm is thus demonstrated.