Electric Vehicle Load Disaggregation Based on Limited Activation Matching Pursuits

Abstract This paper proposes a novel limited activation matching pursuits (LAMP) method to monitor the number of eclectic vehicles (EVs) connected at a charging station and their charging activities. LAMP is able to reflect the two-dimensional characteristics of the number and charging time of the EVs. Constraints are entered on the activation coefficients of the matching pursuits to avoid over matching a particular type of EV. The method includes the development of the basis based on typical EV charging profiles and the improvement of matching pursuits. A case study is undertaken based on real EV data collected from London. The results show that 79.35% EVs can be accurately detected in charging station load profile duration of one week. The number of EVs connected within each half hour can be identified with a relative mean absolute deviation (RMAD) of 21.18%.