Traffic flow prediction method and device by employing spatial and temporal distribution features

Abstract

The present invention relates to the field of traffic monitoring, especially to a traffic flow prediction method and device by employing spatial and temporal distribution features. The method and thedevice convert a GPS track of a vehicle within a period of time to multi-frame image data by taking preset time as an interval, input the multi-frame image data into a convolutional neural network, combine a residual network to obtain integrated traffic flow distribution within a period of time, perform prediction of a normal distribution model for the traffic flow distribution and output a normaldistribution model prediction result. The traffic flow prediction method and device combine the spatial and temporal distribution features to avoid large deviation at the aspect of the prediction result, increase the credibility of the prediction result, fully employ the spatial-and-temporal traffic information, predict the traffic condition of a city within a period of time in future through thecity historical traffic flow information and real-time area traffic flow information so as to better meet an intelligent traffic system and provide demands for traffic information service, traffic control and guidance.

Description

technical field [0001] The invention relates to the field of traffic monitoring, in particular to a method and device for predicting traffic flow using the characteristics of time-space distribution. Background technique [0002] Faced with the increasingly severe traffic pressure, countries around the world have adopted various countermeasures, but the urban traffic problem has not been well resolved. Although many cities in China have begun to implement intelligent services, most of the current traffic platforms initially set up only release and reproduce historical traffic information. For travelers and managers, it is necessary to grasp the changes in traffic conditions. Trend, change from passive to active. Therefore, it is of great significance to the intelligent traffic system to make reliable prediction and judgment of the traffic state information of the road network in advance by using scientific and effective methods. [0003] As an emerging discipline of machin...

AI Technical Summary

Problems solved by technology

[0002] Faced with the increasingly severe traffic pressure, countries around the world have adopted various countermeasures, but the urban traffic problem has not been well resolved

Most neural network methods use backpropagation (BackPropagation, BP). This BP network based on empirical risk minimization has the disadvantage that it is easy to fall into a local optimal solution.

Images