基于深度强化学习的停车位智能动态分配方法

推行停车系统的“用户-停车位”动态匹配是解决“找车位难”、低效寻泊和系统周转率低等问题的有效途径。研究将停车场管理者抽象为智能体,以停车场的时、空、电等环境信息作为状态空间,以是否采取延时匹配和分配的车位编号结合为动作空间,提出基于DQN深度强化学习算法的停车位智能动态分配方法。研究结果表明,研究可有效提高用户寻泊效率、充电需求满足率和停车系统的周转率,且智能延时匹配策略能为用户提供更优质的停车位资源。

Carrying out the"user-parking space"dynamic matching of parking system is an effective way to solve the problems of"difficulty in finding parking space",inefficient parking and low system turnover.This study abstracts the parking lot manager as an agent,takes the time,space,electricity and other environmental information of the parking lot as the state space,and takes the combination of delay matching and assigned parking space as the action space,and proposes an smart dynamic parking space allocation method based on DQN deep reinforcement learning algorithm.The research results show that this study can effectively improve the user berthing efficiency,the charging demand satisfaction rate and the turnover rate of the parking system,and the smart time-delay matching strategy can provide users with better parking space resources.