One Step Ahead Energy Load Forecasting: A Multi-model approach utilizing Machine and Deep Learning
Aristeidis Mystakidis, Evangelia Ntozi, Konstantinos Afentoulis, Paraskevas Koukaras, Γεώργιος Γιαννόπουλος, Napoleon Bezas, Paschalis A. Gkaidatzis, Dimosthenis Ioannidis, Christos Tjortjis, Dimitrios Tzovaras
Abstract
Emerging Energy Load Forecasting (ELF) methodologies assist Distribution System Operators (DSOs) and Aggregators. Energy imbalance among consumption and generation could also be managed with high prediction accuracy, as well as smart grid applications, like Demand Response (DR) events. This study aims to test several algorithms as a solution for ELF. The proposed methodology utilizes machine/deep learning models for time-series forecasting in the domain of energy consumption. Via result comparison it has been illustrated that Neural Networks (NNs), both artificial NNs such as Multilayer Perceptron (MLP) and Long Short-Term Memory (LSTM) recurrent NNs with Extreme Gradient Boosting (XGBoost) were the more accurate ones among other models, showcasing Mean Absolute Error (MAE), R-squared (R <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> ), Root Mean Squared Error (RMSE) and Coefficient Variation of Root Mean Squared Error (CVRMSE) values equal to 1.281, 0.98, 2.238 and 0.147, respectively.