Abstract:

A judgment model of range anxiety degree was proposed based on user behavior of big data of Internet of Vehicles to investigate the influence of range anxiety on user behavior of pure Electric Vehicle (EV). Kernal K-means clustering algorithm was used to analyze the different performance of users’ range anxiety differences in charging frequency, start-stop state of charging (SOC), extremely use behavior, etc. Logical regression algorithm was used to establish the classification and recognition model of range anxiety. The probability of anxiety output by logistic regression model was converted into range anxiety grade by scoring card method. The accuracy and universality of the model were verified via questionnaire. The results show that the prediction accuracy of the model is 95.6%. Therefore, this model can effectively determine the extent of users’ range anxiety, and can be used for big data portrait analysis of EV users if integrated with other analytical dimensions.

Key words: electric vehicle (EV), driver’s range anxiety, big data analysis, use behavior, state of charge (SOC), clustering algorithm