论文部分内容阅读
为了分析交通流的速度不均匀性,提出了一种将聚类分析和概率分布函数拟合相结合的新方法.首先,为了确定最优的子类数,采用两步聚类法对实际的速度数据进行聚类分析,分析表明将速度数据分为2类最能反映交通流的固有类型.然后,将此信息用于指导概率分布函数拟合,采用正态分布、偏正态分布和偏-T分布函数分别拟合各子类数据的概率分布,发现偏-T分布函数拟合精度最高,偏正态分布次之,正态分布最差.模型分析结果表明,所提出的混合分布模型比传统单个分布模型具有更好的拟合能力和通用性.此外,新方法在数据拟合方面更加灵活,且能提供更精确的速度分布模型曲线.
In order to analyze the velocity heterogeneity of traffic flow, a new method of combining cluster analysis and probability distribution function fitting is proposed.Firstly, in order to determine the optimal sub-class number, a two-step clustering method is used to evaluate the actual Speed data are analyzed by cluster analysis, which shows that the velocity data can be divided into two categories which can reflect the traffic flow’s most natural type.Then, this information is used to guide the probability distribution function fitting, using normal distribution, partial normal distribution and partial -T distribution function were fitted to the probability distribution of each sub-type data, found that partial-T distribution function fitting accuracy highest, partial normal distribution followed by the worst normal distribution.Model analysis showed that the proposed hybrid distribution model Compared with the traditional single distribution model has better fitting ability and versatility.In addition, the new method is more flexible in data fitting, and can provide more accurate velocity distribution model curve.