基于聚类非参数检验的动态停车需求分布特征

Dynamic parking demand distribution character based on clustering non-parameter tests

为了探索动态停车需求的分布特征,提出聚类非参数检验分析方法.该方法按照个体-组-类的顺序对多时段随机变量进行聚合,在各聚合阶段运用相应的非参数检验方法校核结果,在确保拟合精度的同时,采用相应的模型描述动态随机变量的分布特征.结合该方法的运用,从问题确定、数据调查、数据处理与分析以及分布特征的归纳总结等环节详细阐述了动态停车需求分布特征的分析流程.以南京市新世界中心为例予以应用,对该建筑配建停车场工作日的动态停车需求分布特征进行了分析,基于拟合得到的10种泊松分布模型和7种伽马分布模型,对分布特征的变化特性进行了归纳总结.

To explore the distribution characters of dynamic parking demands, an analysis method of clustering non-parameter tests is proposed. Multi-period random variables are aggregated according to the sequence of single, group, cluster, and relative non-parameter tests are used during different aggregation phases to validate results, with an aim to obtain models of proper amount to describe dynamic random variables＇ distribution character and assure the fitting precision simultaneously. Based on the method, the research process of dynamic parking demands＇ distribution character is introduced from four aspects such as definition of the problem, collection of data, treatment and analysis of data, and conclusion of the distribution characters. The New World Center in Nanjing is taken as a case study. The dynamic parking demands＇ distribution characters of the accessory parking lot dur- ing workday is effectively explored and the variation of the distribution characters is concluded based on the acquired ten Poisson distribution models and seven Gamma distribution models.