摘要
短时交通流预测是目前智能交通领域的研究热点,文中从实际应用的角度出发,提出了用于流量和速度预测的组合预测模型.该模型包含傅里叶历史估计模型、自回归模型和邻域回归模型三个子模型.详细介绍了组合预测模型的预测机理、模型细节以及用以实现模型实时更新的迭代回归算法.该模型被实际应用到北京市道路预测预报系统中,实际预测误差不超过15%.
Short-term traffic flow forecasting is a research focus of Intelligent Transportation System （ITS）.From a practical view,a combined forecast model is studied in this paper,which includes Discrete Fourier Transform （DFT）model,autoregressive model and neighbourhood regression model. Forecast mechanism and specification are discussed in the paper in detail.In order to update forecast model real-timely,recursive regression method is used to change weights and coefficients of sub-models.This model has been applied to Beijing Traffic Forecast System and the average relative error of prediction is less than 15%in practice.
出处
《武汉理工大学学报（交通科学与工程版）》
2010年第5期874-876,881,共4页
Journal of Wuhan University of Technology（Transportation Science & Engineering）
基金
北京市科委绿色通道项目资助(批准号:D07020601400705)
关键词
组合预测模型
邻域回归
递归回归方法
离散傅立叶变换
combined forecast model
neighborhood regression
recursive regression method
discrete fourier transform
