基于深度强化学习的炼钢车间天车调度方法

针对炼钢车间天车任务产生的动态不确定性,提出了基于深度强化学习算法的炼钢车间天车调度方法。首先,基于强化学习将天车调度问题转化为对天车操作动作序列的求解,采用DQN(Deep Q-network)算法构建动作价值网络模型进行求解;然后,以某钢厂出钢跨天车调度为研究对象,以任务完成总时间最短为目标,介绍了基于深度强化学习的天车调度方法的具体设计;最后,采用实际数据对天车动作价值网络模型进行训练,与目前现场广泛使用的基于固定分区的天车调度方案进行仿真试验对比。结果表明,基于深度强化学习的天车调度方法在任务完成总时间上减少了11.52%,提高了天车任务的完成效率,说明了方法的可行性和优化性,为天车调度研究提供了新的思路。

Crane scheduling method in steelmaking workshop based on deep reinforcement learning

Aiming at the dynamic uncertainty caused by overhead crane operation in steelmaking workshop, a crane scheduling method in steelmaking workshop based on deep reinforcement learning algorithm is proposed.Firstly, based on reinforcement learning, the crane scheduling problem is transformed into solving the sequence of crane operation, and DQN (Deep Q-network) algorithm is used to build the action value network model for solving.Then, taking a steel plant as the research object, taking the shortest time to complete the task as the goal, the specific design of the crane scheduling method based on deep reinforcement learning is introduced.Finally, the actual data is used to train the crane action value network model, and the method proposed in this paper is compared with the current widely used crane scheduling method based on fixed partition by simulation experiments. The results show that the cranescheduling method based on deep reinforcement learning reduces the total task completion time by 11.52%, improves the completion efficiency of the crane task, and proves the feasibility of the method. It provides a new idea for the research of the crane scheduling.