Ensemble Learning for Charging Load Forecasting of Electric Vehicle Charging Stations
Xingshuai Huang, Di Wu, Benoît Boulet
Abstract
Electric vehicles (EVs) can help reduce the dependency on fossil oil and increasing concerns on environmental pollution problems. However, due to the complex charging behaviors and the large charging demand, EV charging has imposed a large burden on the power system. The forecasting of electric vehicle charging loads can help address the above issues by providing power systems with the future load as a reference for energy dispatching. Machine learning methods have demonstrated their effectiveness for short-term load forecasting. Different from previous works, this paper proposes a novel ensemble learning-based forecasting model by combining three base learners including the artificial neural network (ANN), recurrent neural network (RNN), and long short-term memory (LSTM) algorithms. Specifically, a linear regression (LR) algorithm is used to learn the weight of each base learner. The feasibility and advantage of our proposed model are demonstrated by experiments conducted on a real-world dataset and comparisons with the other four baselines.