ALENCAR, David Barbosa de2024-08-052024-08-052021-07-27PARENTE, Ricardo Silva. Modelo híbrido utilizando holt-winters e rede neural não linear autoregressiva com entradas exógenas (narx) para previsão da velocidade do vento. 2021. p. 120. Dissertação do Programa de Pós-Graduação em Engenharia, Gestão de Processos, Sistemas e Ambiental (PPGEGPSA), Instituto de Tecnologia e Educação Galileo da Amazônia (ITEGAM), Manaus, 2021.https://rigalileo.itegam.org.br/handle/123456789/26The wind energy matrix is gradually increasing in recent years and its importance for the renewable energy industry is increasingly linked to the benefits in relation to the environment, with this growing energy matrix the research around the wind power generation is also increasing, and one of the strands is the wind speed prediction, because with this it is possible to predict the wind power generation and decrease the error rate in decision making in the industry of electricity generation through wind power matrix. Considering the problem of decision making and the unpredictability of wind speed, this paper aims to develop a hybrid model for wind speed prediction and consequently wind power generation, based on HoltWinters Exponential Smoothing and Nonlinear Auto-Regressive Neural Network with Exogenous Inputs (NARX). In the materials and methods, the database of the SONDA project (System of National Organization of Environmental Data) organized by INPE (National Institute for Space Research) was used, in which it was chosen to use the anemometric data from the station of Brasilia - BRB and Petrolina - PTR, where data from the years February 2005 to March 2019 of the BRB station were used for training, validation and testing, and from January 2006 to December 2015 of the PTR station for simulations of the HW, NARX and proposed model. The results obtained with the proposed hybrid HW-NARX model were compared with the HW and NARX seasonal time series forecasting algorithms, in which the proposed model was able to achieve better performance and predictability results than HW and NARX for the ultra-short-term, medium-term, and long-term time horizons.pdf.Previsão da velocidade do ventoModelo híbridoHolt-WintersEnergia eólicaRede neural artificialDissertação1.03.00.00-7 Ciência da Computação