Multi-Section Short-Term Traffic Flow Prediction Based on Multi-Dimensional Time Series

Time series has already been widely utilized in the field of traffic flow prediction from the viewpoints of both theoretical exploration and practical implementation; however, the current research mostly limited to the one-dimensional time series and single section predictions. The objective of this paper is to use the multi-dimensional time series to forecast the multi-section short-term traffic flow. The correlation analysis of the multi-section traffic flow parameters (e.g. traffic volume) is firstly conducted to identify several sections with high correlation. Then, these road sections are treated as a whole to perform the threshold regression multi-dimensional time series short-term traffic flow forecasting. Finally, Comparative analysis between the multi-section and single section (ARTMA model) scenarios is presented to demonstrate the efficacy and robustness of the proposed methodology. Computational results indicate that applying the multi-dimensional time series to forecast the multi-section short-term traffic flow is more precise and robust than the single section approach.