������Ȧ���ݵĸ��ٹ�·OD����Ԥ�������編

以英国M42高速公路线圈交通数据为输入,建立预测模型,提出神经网络OD矩阵预测方法. 将预测结果与该公路3a号至7号交叉口间各支路上自动车牌识别装置测得的车辆数据进行对比,验证其有效性. 解决了以下关键问题：利用线圈数据实现OD矩阵预测的可操作性,该类数据的特殊性是否影响模型构建,通过变异数稳定数据转换能否改善模型性能,能否同时进行单个OD预测. 得到基于训练数据平方根代换的最佳计算结果和单个OD预测模型.     

Empirical assessment of urban traffic congestion

This paper presents an empirical assessment of urban traffic congestion in Central London, UK. Compared with freeways or motorways, urban networks are relatively less studied because of its complexity and availability of required traffic data. This paper introduces the use of automatic number plate recognition technology to analyze the characteristic of urban traffic congestion in Central London. We also present the use of linear regression to diagnose the observed congestion and attribute them to different causes. In particular, we distinguish the observed congestion into two main components: one due to recurrent factors and the other due to nonrecurrent factors. The methodologies are illustrated through a case study of Central London Area. It is found that about 15% of the observed congestion in the region is due to nonrecurrent factors such as accidents, roadwork, special events, and strikes. Given the significance of London, the study will be valuable for transport policy evaluation and appraisal in other global cities. Copyright © 2013 John Wiley & Sons, Ltd.