A data-driven two-level clustering model for driving pattern analysis of electric vehicles and a case study

The driving patterns of the electric vehicles describe how users use their vehicles, reflecting the users' habits. These patterns have a positive effect on vehicle energy consumption. In this paper, a two-level clustering model is proposed to determine the driving patterns of electric vehicles. Firstly, the driving pattern characteristics are extracted from the data set of the electric vehicles. Then, the driving patterns including daily driving patterns and multifaceted driving patterns are obtained by a two-level clustering model. The data of 1463 electric vehicles in China were collected from September 1, 2015, to September 1, 2016. Using the proposed model, we obtain five types of daily driving patterns and four types of multifaceted driving patterns. Then, the features of clusters are extracted, and the geographical distribution analysis of the multifaceted driving patterns is conducted. The experimental results reveal that there are many driving patterns of the electric vehicles. Moreover, the effectiveness of the clustering models is verified by the experiments. The customer segmentation based on the driving patterns of electric vehicles is of a great significance for the development of personalized and targeted marketing strategies of vehicle manufacturers and energy efficiency improvement.