Combined Short-Term Traffic Flow Forecast Model for Traffic Forecast System of Beijing Olympics 2008
Date Issued
2009-01
Date
2009-01
Author(s)
Sun, Li Guang
Dong, Shen
Chang, Tang Hsien
Lu, Huapu
Abstract
This paper aims to develop a short-term traffic forecasting model for Beijing, the Olympic city, 2008. From a practical view, a combined forecast model is considered, which includes Discrete Fourier Transform (DFT) model, autoregressive model and neighborhood regression model. In order to update weight real-timely, the Bayesian approach is utilized to activate the weight of each sub-model. The proposed model has been being practiced for the Beijing Traffic Forecast System. According to the study, the average relative error of prediction is less than 15%.
Subjects
Bayes' theorem
Fourier transforms
Operations
Real time control
Real time information
Traffic flow
Traffic surveillance
Description
The TRB 88th Annual Meeting, Washington D.C., USA, Jan.11-15, 2009
Type
conference paper