Risk driving behaviors assessment for freight vehicles based on the Internet of Vehicles data

Driving style is an important factor leading to frequent cargo transportation accidents and serious adverse consequences. To explore the impact of the driving style of freight vehicles on driving safety, the driving data of ten freight vehicle drivers over the course of 31 days was extracted through the Internet of Vehicles (IoV) platform. The risk entropy of acceleration is used to classify the driving style of the driver, the K-medoids method is used to cluster the driving style, and the silhouette coefficient is used to evaluate the clustering quality. We calculate five driving characteristic parameters, and convert them into a few comprehensive variables containing clear driving behavior information based on factor analysis, so as to calculate the risk of drivers with different driving styles. The results show that driving styles fall into three clusters: aggressive, normal, and calm. The driving risk of aggressive drivers is the highest (0.514), and the driving risk of normal drivers (-0.26) is greater than that of calm drivers (-1.54). The research results have significance for the education and supervision of freight vehicle drivers.