Expressway cross-sectional flow prediction algorithm based on spatio-temporal data fusion

Abstract

The short-term traffic flow prediction is of practical significance for the operation and management of the expressway. Combining the traffic flow data collected from the expressway network in Shandong Province, the short-term traffic flow prediction of expressway sections is studied to provide support for the establishment of vehicles travel path selection model. Firstly, the spatio-temporal correlation of traffic flow is analyzed and the correlation coefficients are calculated. Secondly, the summation autoregression and average long-term and short-term memory models based on time series, and regression prediction model based on spatial correlation are selected to predict the traffic flow at the section. Finally, the weighted least squares method is used for the spatio-temporal data integration prediction. The prediction results show that the prediction accuracy of the spatio-temporal traffic flow data fusion algorithm is higher than that of the single prediction model. The average absolute percentage error of the data fusion algorithm is reduced to 8.127%, and the average absolute error and root mean square error are lower than those of the single prediction model. The combined prediction model improves the prediction accuracy.