Title: Truck traffic state prediction based on small sampled location data

Authors: Yiran Ding; Xingkun Li; Xiaozhi Li; Xiucai Zhang; Yuhai Wang

Addresses: College of Automotive Engineering, Jilin University, Changchun, Jilin, 130025, China ' School of Vehicle and Mobility China, Tsinghua University, Beijing, 100084, China ' College of Automotive Engineering, Jilin University, Changchun, Jilin, 130025, China ' College of Automotive Engineering, Jilin University, Changchun, Jilin, 130025, China ' College of Automotive Engineering, Jilin University, Changchun, Jilin, 130025, China

Abstract: Regional road network truck traffic state prediction can provide real-time traffic information for drivers or intelligent devices, which is of great significance for improving road traffic capacity. In this paper, a new method of truck traffic state prediction based on spatio-temporal location data of the Internet of Vehicles (IoVs), ACST, is proposed to match the grid of the spatial road network and form a prediction framework integrating the interaction between roads. The framework is composed of ConvLSTM coding and ResNet structure, and road network attention mechanism is introduced to extract the interaction between roads and sections. Based on the spatio-temporal location data of small sampled trucks, the global traffic state is predicted, and the traffic state on a certain road is focused. Experimental results based on real road dataset show that the results are better than general advanced baseline method.

Keywords: traffic prediction; ConvLSTM; attention; truck IoV; spatiotemporal location data.

DOI: 10.1504/IJHVS.2024.141417

International Journal of Heavy Vehicle Systems, 2024 Vol.31 No.5, pp.604 - 621

Received: 31 Dec 2022
Accepted: 17 Apr 2023
Published online: 12 Sep 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article