Multi-objective FJSP Based on Multi-agent Reinforcement Learning Algorithm

Flexible job shop scheduling problem is the core problem in the actual production of the intelligent manufacturing industry. To solve the problem of process sequencing and machine scheduling in multi-objective flexible job shop scheduling, a multi-agent reinforcement learning (MARL) algorithm is proposed. Shortening the maximum completion time and reducing machine load as the optimization objectives, the objective function is transformed into a problem that can be solved by reinforcement learning. The state, action, and reward functions are established. The Q-learning method is introduced to propose a multi-agent reinforcement learning optimization algorithm. The algorithm is applied to the Brandimarte benchmark example for simulation verification. Compared with other intelligent algorithms, the algorithm has a faster convergence speed and higher utilization rate of the processing machine. Using this algorithm to solve the MK01 example, the minimum maximum completion time is 40. The feasibility, accuracy and efficiency of the intelligent optimization algorithm proposed in this paper are verified.