VHSSA Model for Predicting Short-Term Traffic Flow of Urban Road

Abstract

A short-term traffic prediction model VHSSA (prediction model based on vertical and horizontal sequence similarity algorithm) is proposed for accurate traffic prediction and dynamic route planning. The model is based on the similarity of vertical and horizontal sequences and analyzes historical traffic time series data and cyclic similarity characteristics of traffic volume in urban roads. The model can overcome the deficiency of the traditional model VSSA (prediction model based on vertical sequence similarity algorithm), which focuses only on vertical cyclic sequence similarity. Complete data are transformed into basic sequences that reflect basic characteristics and fluctuant sequences as well as variation characteristics by utilizing the wavelet transformation function. This transformation can achieve both basic and complete sequence prediction. For complete sequence prediction, the paper corrected the fluctuant sequence based on confidence interval and overlapped it with the basic sequence. Verification experiments are conducted to compare the basic and complete sequences of VHSSA and VSSA. Results show that VHSSA prediction is better than VSSA prediction, and the error probability of VHSSA is lower than that of VSSA; the prediction error can meet the actual requirement.