Abstract: Beijing and other places get car indicators by random shaking way currently. However simple random manner will lead to “shake not” phenomenon, and not consider the rigid demand for vehicles. For the above unfairness, a weighted sampling strategies by groups is presented in this paper：Group model based on shaking time, considering unshaken numbers, eliminate?“shake not” by assigning group rates；Group model based on the number of vehicles, considering the actual needs, focus on some applicants?by assigning the extraction ratio. Combined this two models, this paper proposes vehicle indicators randomly distributed algorithm based on multi-sampling strategy, eliminating the unfairness comprehensively. The simulation results show that the algorithm presented in this article can effectively reduce the unreasonable factors, providing a new way to solve the vehicle?shaking problem.

Key words: random generator, latin hypercube sampling, license rocking, multi-sampling strategy

摘要： 北京等地目前采用随机摇号方式取得购车指标。简单的随机方式会导致“多摇不中”现象，也未考虑申请人对车辆的刚性需求。针对上述不合理性，提出了通过分组进行带有权重的抽样的策略：基于摇号次数的分组模型，考虑未摇中次数，通过分配组间比率消除“多摇不中”；基于家庭车辆数的分组模型，考虑申请人实际需求，通过分配抽取比例对申请人进行侧重。结合两种模型，提出了基于多重采样策略的车辆指标随机派发算法来综合消除不合理性。仿真实验表明，提出的算法能够有效地减少摇号过程中的不合理因素，为车辆摇号问题提供了一种新的解决途径。

关键词: 随机数生成算法, 拉丁超立方抽样, 车牌摇号, 多重采样策略