Prediction for recognition probability of traffic information at intersection of interchanges

Abstract:

To predict the recognition probability of traffic information at the intersection of interchanges, an approach using vehicle dynamics theory, driver characteristic principle, and dynamic traffic characteristics was adopted to obtain the recognition distance. On the basis of this approach, geometric and statistical principles were applied to establish a framework of traffic information within the recognition distance of typical automobiles. Furthermore, the traffic volume at the intersection of interchanges was predicted using short-and long-term time series, and the prediction results were subsequently implemented to form a prediction model for the recognition probability of traffic signs, which was validated with actual measurement. Results show that under long-term time series prediction, the traffic volume had a significant correlation with the traffic information recognition probability, and the correlation coefficient was 0.849. As for the short-term time series within a week, the overlapped area of 95% prediction interval band between the predicted value and the measured value reached 87.65%, which signifies a high reliability of the prediction model. Based on the high probability of traffic information recognition problems, for the intersection of interchanges with large traffic volume, consideration should be given to strengthen flexible traffic information settings and traffic control measures.