摘要
汽车零件生产车间一直是制造业中的关键环节之一,为了适应动态多变的汽车零件生产车间物流搬运场景、减少自动导引车(AGV)为完成任务所行驶的路程以及发生碰撞冲突的次数从而达到节能目的,提出了一种基于深度强化学习的AGV调度算法。首先,对汽车零件生产车间AGV调度问题进行建模并明确约束条件和目标函数,将车间地图栅格化构建算法训练环境。其次,建立基于深度Q网络DQN的AGV调度算法,设计其状态空间、动作空间、奖励函数。最后,在不同问题场景下通过实验证明基于Double DQN的AGV调度算法相较于传统调度算法的节能优势。
Auto parts production workshop has always been one of the key links in the manufacturing industry.In order to adapt to the dynamic and changing logistics handling scene of auto parts production workshop,reduce the distance traveled by AGV(automated guided vehicle)to complete the task and the number of collisions,so as to achieve the purpose of energy saving.An AGV scheduling algorithm based on deep reinforcement learning is proposed.Firstly,the AGV scheduling problem of auto parts production workshop was modeled and the constraint conditions and objective function were defined,and the workshop map was rasterized to build the algorithm training environment.Secondly,the AGV scheduling algorithm based on Double DQN(Deep Q Network)is established,and its state space,action space and reward function are designed.Finally,the experiment proves the energy-saving advantage of AGV scheduling algorithm based on Double DQN compared with traditional scheduling algorithm in different problem scenarios.
作者
荣子鸣
Rong Ziming(Business School,Xinjiang University,Urumqi 830000,China)
出处
《现代计算机》
2024年第5期81-86,111,共7页
Modern Computer
关键词
自动导引车
调度
节能
深度强化学习
automatic guided vehicle
scheduling
energy saving
deep reinforcement learning
