Determining the Occupancy Patterns and Typical Power Demand Profiles of the Electric Vehicle Charging Stations Using a Flexible Clustering-based Data Mining Framework

Due to the increasing number of electric vehicle charging events, an enormous data volume can be generated and recorded in very large databases. A deeper analysis of the data collected by the monitoring systems could provide significant information to characterize the behavior of the electric vehicle charging stations. Unfortunately, the features of occupancy patterns or the typical power demand profiles corresponding to the charging stations are many and various and can be hard to identify and interpret without adequate techniques. In this context, a flexible clustering-based data mining framework has been proposed to extract the main features (the number of hours when there is at least one charging, number of electric vehicles, and total energy consumption) associated with the occupancy patterns of the charging stations and to determine the typical electric vehicle charging demand profiles. The efficiency of the proposed framework has been demonstrated using a database containing information from the monitoring systems of CSs installed in a city in Spain.