Driving Style Classification Method Based on High-frequency Data from Pure Electric Vehicles

Abstract:

15 characteristic parameters related to driving safety were selected based on the operation data of pure electric vehicles. According to the statistics of the parameters， threshold line with multiple parameters was proposed. Then two clustering algorithms and four dimension-reduction methods were compared. Based on the classification effect， the combination algorithm of t-Distribution stochastic neighbor embedding （t-SNE） and Gaussian Mixture Model （GMM） were used to establish the driving style classification model， which was used to classify the driving style into three classes. On this basis， the vehicle operation characteristics of the three classes were compared. The classification model can effectively reflect the driving behavior and provide reliable evaluation basis for fleet management and road safety.