Hierarchical High-Resolution Load Forecasting for Electric Vehicle Charging: A Deep Learning Approach
Zhile Yang, Tianyu Hu, Juncheng Zhu, Wen‐Long Shang, Yuanjun Guo, Aoife Foley
Abstract
The mass roll-out of electric vehicles substantially contributes to reducing fossil fuel consumption and environmental emissions. Meanwhile, the large fluctuation, strong uncertainty, and multiple coupled impact factors of plug-in electric vehicle charging significantly challenge the existing load forecasting approaches. Unlike the conventional load, the fluctuation of plug-in electric vehicle charging is particularly significant at the very-short-term minute-level time scale given the large charging power and stochastic human behaviors. In this article, a novel historical-data-based high-resolution plug-in electric vehicle load forecasting model structure is established, accommodating the various factors that affect the charging load. Moreover, an enhanced attention-based long short-term memory deep learning approach is proposed for solving the intractable forecasting problem, where a feature upscaling and downscaling algorithm is proposed for the hierarchical high-resolution data processing. Numerical results on real-world data of a charging station in Shenzhen show that the proposed model structure and algorithm demonstrate well performance in short-term and very-short-term hierarchical high-resolution plug-in electric vehicles charging load forecasting problems.