Intelligent guidance method for continuous vehicle dynamic path in connected internet of vehicles environment

This study proposed an intelligent guidance method for continuous vehicle dynamic path in connected Internet of Vehicles environment. Utilizing Internet of Vehicles technology to collect multi-source heterogeneous traffic flow data, high-precision filling of missing data is achieved through self attention - generative adversarial network (SA-GAN). Adopting the spatio temporal graph model predictive control (STGMPC) framework for regional traffic flow prediction. Based on the predicted results, combined with the K-shortest path algorithm, multiple continuous vehicle dynamic paths are generated and recommended for online rolling of vehicles, achieving intelligent vehicle guidance. The experimental results show that the maximum utilization rate of spatiotemporal resources of the proposed method is 92.3%, the maximum guidance success rate is 98.2%, and the minimum guidance instruction response time is 0.4s.