A Bi-Level Method for Flexibility Feature Extraction and User Clustering Based on Real-World Data from Independent Smart Meters of Residential Electric Vehicle Users

Residential electric vehicle (EV) chargers in China are typically paired with independent smart meters. A bi-level method for extracting flexibility features and clustering residential EV users is proposed based on real-world smart meter data, offering significant advantages in terms of data accessibility and implementation feasibility. First, real-world smart meter data from over 5000 residential EV users are partitioned into charging segments, which are then filtered and grouped. At the charging segment level, three charging features are extracted using the Pearson correlation method, and six categories of charging segments are clustered using a Gaussian Mixture Model (GMM). Subsequently, at the user level, two flexibility features are extracted based on the charging segment clustering results, and five categories of residential EV users are clustered using the K-means algorithm. The results of the bi-level method are presented and analyzed, with its effectiveness validated by comparing the trends of flexibility features across different user categories between the training dataset and the validation dataset. In the concluding section, the limitations of the current research are discussed, and potential directions for further research are outlined.