Short-time traffic flow prediction method considering diffusion process

Abstract

The invention discloses a short-time traffic flow prediction method considering a diffusion process, the method comprises the following steps: obtaining a historical traffic flow sequence of the current road section and performing a smoothing operation to obtain a smoothed traffic flow sequence F = {X1, X2 - X<t-1>}; adopting an LSTM-CNN model to capture a depth spatial-temporal feature from the smoothed traffic flow sequence; performing a diffusion process of digital description through a class PageRank algorithm to obtain a road importance feature of the current road section from the smoothed traffic flow sequence, wherein the road importance feature and road auxiliary information are combined to form a one-dimensional vector serving as a breadth feature; and fusing the depth spatial-temporal feature and the breadth feature to obtain a traffic flow prediction value Xt of the current road section at the t moment.

Description

technical field [0001] The invention relates to a short-term traffic flow prediction method considering the diffusion process, which belongs to the cross field of intelligent traffic, computer and mathematics. Background technique [0002] With the rapid development of social economy and the acceleration of urbanization, effective traffic control and guidance has become an urgent problem to be solved, and real-time and accurate traffic flow forecasting is a crucial step. However, the current research does not fully consider the dynamic changes of traffic flow on the whole road network, resulting in the prediction accuracy not meeting the requirements. Contents of the invention [0003] The main purpose of the present invention is to overcome the deficiencies of the prior art, to propose a short-term traffic flow prediction method considering the diffusion process, by adding features such as road importance, to carry out accurate short-term traffic flow prediction, to solve...

AI Technical Summary

Problems solved by technology

[0003] The main purpose of the present invention is to overcome the deficiencies of the prior art, to propose a short-term traffic flow prediction method considering the diffusion process, by adding features such as road importance, to carry out accurate short-term traffic flow prediction, to solve the existing prediction The method considers the problem of insufficient comprehensiveness in terms of dynamic globality, and further improves the prediction accuracy

Images