Deep Translation Model-Based Electric Vehicle Load Disaggregation Method

With the rapid development of electric vehicles (EVs), energy systems and grid operations are facing new challenges and opportunities. As a key component of renewable energy and smart grids, managing EV loads holds significant importance in allocating electrical loads sensibly and minimizing the impact of uncontrolled vehicle charging on the grid. Obtaining detailed information about EV charging is particularly challenging when network operators lack comprehensive monitoring equipment. Therefore, this paper proposes a sequence translation model-based approach for disaggregating EV loads, enabling the refined acquisition of information on each charging event while saving on sensor costs, thereby aiding in determining users’ charging needs. We design a combined charging mode state code in order to enhance the disaggregation efficiency. Subsequently, by leveraging the theory of sequence translation models to comprehensively extract the features of the data, we establish a mapping model between the overall charging curve and the combined charging mode state code. Finally, we validate the proposed method based on elaadNL dataset, achieving an 80% accuracy rate in load disaggregation, thus confirming the efficacy of the presented approach.