Traffic State Forecasting of Typical Roads in Beijing

The comprehensive analysis of traffic states are effective ways to eliminate some traffic problems and enhance the supervising ability of the traffic management departments, which includes real-time monitoring, evaluation and prediction the states of the entire city. With analyzing the traffic flow characteristic on several typical roads in Beijing and summarizing the existing traffic prediction models, this paper proposes a combination forecasting model. It is mainly based on the nonparametric regression model. The combination of Fourier's history estimated model, nonparametric autoregressive model and nonparametric neighborhood regression model are used to predict the traffic state of typical Roads. Considering the reality of the traffic flow information collection system of Beijing and the needs of predicting information released by graphical way in the future, the paper also presents the heterogeneous data fusion methods and road traffic code model.

摘要：对整个城市的道路交通运行状况进行全面分析评估并实时监控、预测，可以有效地消除交通隐患，增强城市交通管理部门对城市交通的管控能力。本文基于北京市典型道路交通流特性分析及已有的道路交通流预测模型分析，提出了道路交通运行状态组合预测模型，确定了非参数回归模型作为预测模型的核心，组合使用傅立叶历史估算模型、非参数自回归模型和非参数邻域回归模型对北京市典型道路的交通运行状态进行预测。针对北京市道路交通流信息采集系统实际情况及未来预测信息图形化发布的需要，提出了道路交通流预测系统的异构数据融合方法及道路编码模型及方法。