Charging Behavior Analysis Based on BIRCH Clustering

Many charging pile statistics have been produced as a result of the increased popularity of electric vehicles and the ongoing growth in the number of charging stations. In order to obtain a typical charging user profile, this paper collects and cleans the charging data from 2021 to 2022 in Banan District, Chongqing. It then uses the BIRCH clustering method to group the charging power, SOC, and RFM data into one-dimensional, two-dimensional, and three-dimensional cluster groups. According to the clustering results, 75% of users in the Banan District charge at low and medium power levels. Some users exhibit overt signs of anxiety about their mileage or refuse to wait for charging. RFM clustering categorizes the level of user demand for charging in the Banan District into three types, demonstrating how frequently users charge there. Finally, this research offers several recommendations based on the three clustering traits. The user profile and recommendations can successfully aid distribution networks and operators in better understanding users, and they can serve as useful resources for creating better charge configuration plans and marketing campaigns.

I. Introduction

The number and size of EV and charging piles continue to grow with the development of new EVs, creating a significant amount of multi-dimensional structured user data. The planning and operation of the electricity distribution network and operators rely heavily on the analysis and mining of the enormous amounts of big data produced by automotive operation.