A short-term, real-time system was developed to support traffic management in Beijing. The requirements of a large amount of data and unstable traffic flow are the biggest challenges to such a system. The models and software framework thus should be effective enough to face these problems. The core of such a system is the short-term traffic flow forecast model. Rapid urbanization and transportation development in Beijing have led to traffic flow patterns with some unstable characteristics. The short-term forecast model for an online system thus was designed with the fast-paced trend in mind. The model considers historical data, real-time data, and space data, and it can be updated online. Thus a combined model was developed with three submodels: discrete Fourier transform model, autoregressive model, and neighborhood regression model. Weights of each submodel were based on forecast error. Both the historical forecast error and real-time forecast error were considered. The system was built on a browser–server structure to support combined forecast models. The framework, modules, and interface of this system are introduced in this paper.
Short-Term Traffic Forecast System of Beijing
Transportation Research Record
Transportation Research Record: Journal of the Transportation Research Board ; 2193 , 1 ; 116-123
2010-01-01
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch
Short-Term Traffic Forecast System of Beijing
Online Contents | 2010
|LSTM network: a deep learning approach for short‐term traffic forecast
Wiley | 2017
|An Adaptive-Margin Support Vector Regression for Short-Term Traffic Flow Forecast
Taylor & Francis Verlag | 2013
|