Charging infrastructure is essential for promoting the development of electric vehicles, and identifying charging behaviors of electric vehicles is the precondition of optimizing the layout of charging infrastructure. Using the charging data of the Kechuang Base Charging Station in Beijing in 2017, charging behaviors of electric vehicles were explored by descriptive analysis and statistical analysis. Based on the charging power, charging piles were divided into three categories: high (100 kW), medium (40 kW), and low(10 kW and 15 kW). Firstly, a descriptive analysis was conducted. It is found that as the charging power decreases, the charging time significantly increases, but usually does not exceed 180 min. 86.5% of customers are company users, mainly consisted of taxi/ride hailing drivers; electric vehicles are often charged when their state of charge (SOC) are still high. Then, an ordered Logistic model was built to identify the key factors influencing the charging pile choice. Company users, daytime, weekday, charging peak period, and the low starting SOC are found to be able to significantly lead users to adopt the high-power charging piles. The research findings could be used to help optimizing the charging station layout.