基于元胞机与强化学习的多扰动车间调度算法

Scheduling algorithm for multi-disturbance job-shop based on cellular automata and reinforcement learning

针对大型装备制造企业扰动多、影响大的问题,以元胞机模型为框架构建了多扰动车间生产调度模型,设计了基于设备平均利用率与工件平均流程时间双目标最优的目标函数,并通过算例验证了模型的科学性。同时,考虑到多扰动车间调度的复杂性,为寻找全局最优解,采用强化学习算法优化了元胞机的自组织演化规则,提出了针对设备故障、紧急插单与新订单干扰三种典型扰动的调度策略,最终建立了基于元胞机与强化学习算法的多扰动车间柔性调度模型。以某大型零件制造企业为例,说明了模型的具体优化过程,并通过仿真求解验证了算法与模型的有效性与可靠性。

To solve the problem of large equipment manufacturing enterprises with many disturbances and great influence,a multi-disturbance job-shop production scheduling model was built based on the cellular automata,and the objective function based on the average utilization rate of equipment and the average flow time of workpiece was designed.The model was proved to be scientific by introducing an example.In view of the complexity of multi-disturbance job-shop scheduling,the reinforcement learning algorithm was used to optimize the cellular automata evolution rules to find the global optimal scheduling solution,which contributed to design scheduling strategy for three typical disturbances of equipment failure,emergency insert sheet and new orders interference.A multi-disturbance flexible job-shop scheduling model was established based on cellular automata and reinforcement learning algorithm.A large part manufacturing enterprise was taken as an example to illustrate the specific optimization process of the model,and the effectiveness and reliability of the algorithm and model were verified by simulation.