Research on the Application of Smart City Super Brain in the Upgrading and Reconstruction of Expressway—Take Chengdu Old Chuan-Zang Road as an Example

Abstract

This paper explores the smart city super brain traffic flow prediction method. Through monitoring and studying the traffic volume and average speed of related intersections in the morning and evening peak hour and one normal hour, a total of 7413 data are collected, of which 5930 are used as training sets and 1483 are used as test sets. Long short-term memory network (LSTM) method is used to predict the changes of traffic flow at important traffic nodes along old Chuan-Zang Road in Chengdu and the changes of vehicle speed at typical time of one day. The predicted results are highly coincident with the actual data, which proves the accuracy and effectiveness of the model. According to the predicted results, through the simulation analysis of the transformation scheme of important nodes, the traffic operation of the study section before and after the transformation is evaluated. The results show that the average delay of vehicles is reduced by 23.82% compared with that before the transformation, and the pollutant emission and fuel consumption are also significantly reduced. LSTM, with its unique advantages in processing sequence data and solving long-term nonlinear problems, provides algorithms and basis for urban road upgrading.