The vast increase in electric vehicles (EV) in the market for business and private use implies a necessity to adapt and improve the charging infrastructure. One of the main assets that can be deployed to achieve this goal, is the use of the measured data of the EV supply equipment (EVSE). These data can be implemented to provide a coordinated power management method, which allows coordinated charging according to some predefined criterion and results in a reduced load peak. Analyzing the measured data allows to extract certain features from the charging profile of individual vehicles during the actual charging session. These features represent not only the predefined charging process based on the SOC of the battery but also the behavioral characteristic of the user of the vehicle in terms of its intended use. In this work, a novel approach is introduced, comprising a combined machine learning (ML) classification algorithm and charging power optimizer (CPO). The ML algorithm classifies the vehicles that are regular users of the charging ports within a certain building. Classification takes place in in terms of electric charging profile features and behavioral characteristics of the user or the intended use. The CPO enables a smart and coordinated charging process depending on both the EVSE current state and the participating vehicles. The coordinated charging is based on regulating power rates of charging sessions to anticipate the electrical load. As a proof of concept, the FKFS research EVSE is employed to collect the data from a regular group of fleet and private EVs. Then an in-house-made charging test system emulates the action of the CPO. The combined action of this approach provides a smart and coordinated charging, which guarantees proper charging of the vehicles, reduced load peaks of the building and hence avoid grid congestion.
Session: CHARGING | | 09:00 - 09:30