【摘要】 针对求解最小化最长完工时间的作业车间调度问题，提出一种基于虚拟仿真与深度强化学习联合优化的方法，旨在缩小调度问题模型中理想与实际生产车间之间的差距。在Plant Simulation平台搭建仿真车间模型作为强化学习环境，以套接字方式交互反馈奖励值。应用近端策略优化强化学习算法对动作选择策略网络与状态评价网络进行训练。算例表明，该虚拟仿真技术与强化学习的集成调度具有良好的鲁棒性与泛化性，且相较于其他算法优化了调度性能，加快了求解速度。为智能制造工程及相关专业基于问题式学习的实践类课程提供了典型应用案例。更多还原

【Abstract】 In order to bridge the gap between the ideal and the actual production plant in the scheduling problem, an integrated optimization method based on discrete simulation and deep reinforcement learning is proposed to solve the job-shop scheduling problem that is to minimize the maximum completion time. A simulated job-shop model is built using plant simulation platform as the reinforcement learning environment to feedback the reward values via socket. The proximal policy optimization algorithm is applied to train the action selection strategy network and the state evaluation network. Experimental results show that the integrated scheduling of virtual simulation technology and reinforcement learning has good robustness and generalizability. Compared with other algorithms, it shows better scheduling performance with high solving speed. It is a typical application case for practical courses that are problem-based learning in intelligent manufacturing engineering and related majors.更多还原