Reliable Real-Time Charging Profile Estimation for Fast EV Chargers Under Faulty Conditions

This paper proposes a real-time charging profile forecasting method for fast electric vehicle (EV) chargers. The proposed method aims to estimate voltage, current, power, and state-of-charge (SoC) profiles in the event of sensors faults, communication faults, or both. The healthy measurements are initially used to develop a dynamic model. When a fault occurs, this model forecasts the charging profile for the remainder of the session without relying on additional measurements. The proposed method considers the model of the battery as a black box, and utilizes an adaptive recursive least squares (RLS) algorithm to estimate the internal parameters of the battery model. This ensures a reliable reconstruction of missing sensor's data under post-fault conditions. To evaluate the effectiveness of the proposed approach, various mathematical and approximation models are analyzed and compared for accuracy using Matlab.