RT info:eu-repo/semantics/article
T1 Novel data-driven models applied to short-term electric load forecasting
A1 López Martín, Manuel
A1 Sánchez Esguevillas, Antonio Javier
A1 Hernández Callejo, Luis
A1 Arribas Sánchez, Juan Ignacio
A1 Carro Martínez, Belén
K1 Short-term electric load forecasting
K1 Deep learning
K1 Machine learning
K1 Deep learning additive ensemble model
K1 Dynamic mode decomposition
K1 33 Ciencias Tecnológicas
K1 3325 Tecnología de las Telecomunicaciones
AB This work brings together and applies a large representation of the most novel forecastingtechniques, with origins and applications in other fields, to the short-term electric load forecastingproblem. We present a comparison study between different classic machine learning and deeplearning techniques and recent methods for data-driven analysis of dynamical models (dynamicmode decomposition) and deep learning ensemble models applied to short-term load forecasting.This work explores the influence of critical parameters when performing time-series forecasting,such as rolling window length, k-step ahead forecast length, and number/nature of features used tocharacterize the information used as predictors. The deep learning architectures considered include1D/2D convolutional and recurrent neural networks and their combination, Seq2seq with andwithout attention mechanisms, and recent ensemble models based on gradient boosting principles.Three groups of models stand out from the rest according to the forecast scenario: (a) deep learningensemble models for average results, (b) simple linear regression and Seq2seq models for veryshort-term forecasts, and (c) combinations of convolutional/recurrent models and deep learningensemble models for longer-term forecasts.
PB MDPI
SN 2076-3417
YR 2021
FD 2021
LK https://uvadoc.uva.es/handle/10324/54157
UL https://uvadoc.uva.es/handle/10324/54157
LA eng
NO Applied Sciences, 2021, vol. 11, n. 12, p. 5708
NO Producción Científica
DS UVaDOC
RD 19-oct-2024