基于贝叶斯智能学习OD矩阵估计与网络拓扑优化研究

OD矩阵（Origin—Destination Matrix）是路网规划与评价的基础数据。以往OD矩阵数据是通过交通调查的方法获得，这往往耗费了大量的人力和物力。运用贝叶斯定理的先验分布原理，构造贝叶斯智能学习的网络拓扑结构，提出由各路段交通流量的观测值来推算估计以及预测OD矩阵的一种有效方法。利用此方法可以准确估计出OD矩阵数据，同时在优化网络拓扑中，能对未来交通量的分配进行预测。对比分析表明，此方法能有效地提高交通运输规划的效率以及交通评价准确性。

Study on OD Matrix Estimation and Network Topology Optimization Based on Bayesian Intelligent-studying

OD Matrix is the basic data for roadway network planning and evaluation, In the past, the OD Matrix data were got by traffic survey which usually cost too much work, In this paper, an intelligent-studying Network Topology is constructed with the theory of Bayesian statistics. A method for estimating and predicating OD matrix by traffic volume at roadway bl~k is presented. By this method, the OD Matrix data can be evaluated accurately, and the traffic volume can be forecasted in optimizing the intelligent network topology. The contrastive analysis proves that the method can improve the efficiency of the traffic planning and the accuracy of the traffic evaluation.