Abstract: The delayed departure of shared users keeps parking time in conflict. Based on this, the random dynamic programming has been constructed. In this paper, the reservation control policy has been studied, and two algorithms have been designed to solve the problem of whether the stay-over requests can be rejected or not. Finally, the effectiveness of algorithms has been verified by numerical simulation. The conclusions are as follows: (1)As the reservation period increases, the marginal revenue of the platform will increase and then decrease, so there is an optimal value of the reservation period;(2)If the platform can decide whether the stay-over requests is accepted or not, then the opportunity cost will decrease with the increase of the stay-over probability, and the service rate of the platform and the utilization ratio of the parking capacity will increase; (3)If the platform can only passively accept the stay-over requests, then the service rate of the platform will decrease with the increase of the stay-over probability, and the opportunity cost and the utilization ratio of the parking capacity will increase.

Key words: shared parking, random dynamic programming, reservation control policy, stay-over

摘要： 基于车位共享用户未按约定时间离开而导致车位使用时间冲突的问题,构建了随机动态规划模型。通过模型,研究了平台的预订控制策略,并针对平台能否拒绝延时需求的情形,设计两种算法对模型进行求解,最后通过数值仿真验证了算法的有效性。研究结果表明:(1)随着预订周期的增加,平台的边际收益会先增加后减少,因此存在最优预订周期;(2)若平台可以决策延时需求是否被接受,则随着延时概率的增加,机会成本将减少,而平台服务率和车位利用率将增加;(3)若平台只能被动接受延时需求,则随着延时概率的增加,平台服务率将减少,而机会成本和车位利用率将增加。

关键词: 车位共享, 随机动态规划, 预订控制策略, 延时