关键词: 城市交通, 分布预测, 回归分析, OD矩阵, 道路网络, Zipf定律

Abstract:

Traffic distribution data is difficult to obtain and cost high, which is the main factor restrict traffic engineers to carry out traffic prediction. In order to improve the efficiency and reduce the forecasting cost in engineering practice, we studied the predictability of OD matrix in urban road network. In this paper, three kinds of traffic distribution models are described. Taking Guangzhou as an example, the probability distribution of traffic flow and the probability distribution of production are given. By means of regression analysis and residual analysis, the relationship among production, population and economy is explored. We proposed a target dual-factor model considering population benefits (called TDM model). The error analysis is used to verify the accuracy of the four models and then the TDM model is applied in Shenzhen. The results show that the probability distribution of traffic flow and the probability distribution of production are highly heterogeneous and follow Zipf 's law. There is a strong correlation among the production, population and economy, and goodness of fit test is 0.87. The accuracy of the TDM model is slightly lower than that of the gravity model, but higher than that of the other two models. In addition, the result of forecast in Shenzhen is good. Considering prediction accuracy, cost and efficiency, the TDM model is more suitable for predicting urban road traffic distribution.

Key words: urban traffic, distribution prediction, regression analysis, OD estimation, road network, Zipf 's law