This paper models and solves the optimal allocation-pricing of reservable parking resources and the pricing of unreservable parking resources, respectively. For reservable parking facility, a MP-DGS (modified proxy Demange-Gale-Sotomayor) mechanism and combinatorial system (integration of direct and evolutionary methods) are adopted to maximize the social surplus through optimizing the allocation-pricing of parking permits. As a result, it is found that: (i) the proposed approaches not only simplify the users’ bidding procedures but also ensure the users express their preference truthfully even under the situation of non-optimal parking permits allocation; (ii) in homogeneous case (parking periods for all users are the same), it is theoretically demonstrated that the MP-DGS mechanism is more efficient than the traditional mechanisms in the worst-case scenario; (iii) in heterogeneous case (users are heterogeneous in desired parking timing and duration), time-dependent parking permits are taken into account. The ranking of the algorithm time complexity in the worst-case scenario is that direct method = evolutionary method < Leonard mechanism = VCG (Vickrey-Clarke-Groves) mechanism, and the combinatorial system not only solves out the optimal allocation-pricing results effectively but also ensures the optimal results can be obtained in a shorter time. In addition, for unreservable parking facility, we formulate a dynamic social optimum as a stochastic control problem and then obtain a region-based optimal dynamic parking pricing. Through theoretical analysis, it is revealed that depending on the realization of the queue length due to the cruising-for-parking, the region-based optimal dynamic parking pricing can be divided into two patterns, furthermore, each pattern results in a “bang-bang” control.

本文分别对可预留停车资源的最优配置定价和不可预留停车资源的定价进行建模和求解。对于可保留的停车设施，采用MP-DGS（改进的代理人Demange-Gale-Sotomayor）机制和组合系统（直接与进化方法相结合），通过优化停车许可证的分配定价来最大化社会盈余。结果，发现：（i）所提出的方法不仅简化了用户的投标程序，而且即使在分配停车许可证不是最佳的情况下，也能确保用户如实表达自己的偏好；（ii）在同类情况下（所有用户的停车时间都相同），从理论上证明，在最坏的情况下，MP-DGS机制比传统机制更有效；（iii）在异类情况下（用户在所需的停车时间和持续时间上是异类的），应考虑到随时间变化的停车许可证。最坏情况下算法时间复杂度的排序是直接方法=演化方法<Leonard机制= VCG（Vickrey-Clarke-Groves）机制，组合系统不仅可以有效地解决最优分配定价结果，而且还确保可以在更短的时间内获得最佳结果。此外，对于不可保留的停车设施，我们将动态社会最优公式化为随机控制问题，然后获得基于区域的最优动态停车价格。通过理论分析