In this work we will address the short-term electricity consumption forecasting problem related to the electric vehicle load demand. In particular we will focus on the explainability of the model obtained. These are important aspects of this problem, since it would help gaining insight on the most important features involved in the forecasts. For the purpose of forecasting, we will use linear regression and three machine learning methods: random forest, gradient boosting and long short-term memory artificial neural network. Later, We add an explainability layer to the models generated, to get a better understanding of the predictions. As far the predictions are concerned, results obtained by the long short-term memory neural network are more accurate than those obtained by random forest and gradient boost, having used linear regression as baseline. The features that most contribute to the predictions are the 25 closest to the present but also a set of features with 30 to 60 unit lag.
CITATION STYLE
Gallardo-Gómez, J. A., Divina, F., Troncoso, A., & Martínez-Álvarez, F. (2023). Explainable Artificial Intelligence for the Electric Vehicle Load Demand Forecasting Problem. In Lecture Notes in Networks and Systems (Vol. 531 LNNS, pp. 413–422). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-031-18050-7_40
Mendeley helps you to discover research relevant for your work.