The detection and analysis of power quality is the purpose and basis for studying the interaction between electric vehicles and the power grid. However, most existing work has not used real-world data to evaluate the power quality of charging stations. The time analysis scale of the existing measured data research is mostly at minute level, and there is a lack of comparative analysis on the different characteristics of different types of charging stations. Given this, this paper presents a power quality analysis method for an electric vehicle charging station based on measured data. First, through the visualization of the charging platform data, the time sequence characteristics of the actual arrival of charging vehicles at the integrated charging station and the bus charging station are extracted. Secondly, the theoretical basis of the influence of charger topology on harmonic characteristics is established by simulation modeling, and the comparison between simulation data and measured data is given. Then, the measured power quality indices of the above two types of station are qualitatively and quantitatively evaluated in combination with relevant national standards, and their advantages and disadvantages are compared. Finally, based on the measured data, different experimental scenarios are designed to deeply mine the related factors affecting the harmonic characteristics of charging stations. The results show that the harmonic level of an integrated charging station is better than that of a bus charging station. Compared with electric buses, electric private vehicles cause less current harmonics to charging stations. In addition, factors such as the total charging power, harmonic cancellation amplitudes between harmonic sources of multiple charging piles, access and disconnection event of electric vehicles, and different load combinations etc., all have impact on the power quality of the charging station. This approach provides a novel idea for harmonic monitoring and treatment of electric vehicle charging stations. This work is supported by the National Key Research and Development Program of China (No. 2016YFB0900602) and the Science and Technology Project of State Grid Corporation of China (No. SGTJBH00YXJS1902795).