Traffic Flow Velocity Prediction Based on EMD-BP Method

Abstract

With the continuous development of urbanization in China, the urban traffic network is constantly improving, the construction process of smart cities is accelerating, and the prediction of urban traffic road speed is increasingly important as urban road planning, construction and mastering the current road operation status for traffic management departments. By analyzing the historical data of urban road traffic flow for a week, aiming at the characteristics of time-varying, non-linear and non-stationary characteristics of traffic flow, the Smooth-EMD-BP combined forecasting model is constructed to predict the running speed of urban road traffic vehicles. The prediction accuracy of five kinds of evaluation indicators is introduced. The results show that the root mean square error of this method is 0.0049, which is much smaller than the root mean square error of the model established by BP network. The model proposed in this paper is more in line with the urban road traffic flow vehicle running speed.