Avaliação de Topologias de Redes Neurais Artificiais para Previsão de Potência e Energia Fotovoltaica

Gustavo Henrique de Paula Santos; Paul Junior Zapana Vargas; Elmer Pablo Tito Cari; Moisés Carlos Tanca Villanueva

doi:10.20906/CBA2022/3223

Gustavo Henrique de Paula Santos Departamento de Engenharia Elétrica e Computação, Universidade de São Paulo - USP, São Carlos
Paul Junior Zapana Vargas Departamento de Engenharia Elétrica e Computação, Universidade de São Paulo - USP, São Carlos
Elmer Pablo Tito Cari Departamento de Engenharia Elétrica e Computação, Universidade de São Paulo - USP, São Carlos
Moisés Carlos Tanca Villanueva Departamento de Ingeniería Eléctrica, Universidad Nacional de San Agustín de Arequipa - UNSA, Arequipa

DOI: https://doi.org/10.20906/CBA2022/3223

Keywords: Renewable sources, Photovoltaic power, Photovoltaic energy, Power forecasting, PV system, Artificial neural networks

Abstract

Renewable energy sources such as solar photovoltaic have grown every year. Although beneficial, the integration of photovoltaic solar generation systems can cause problems due to meteorological variations that lead to uncertainties in energy production. In this sense, this work evaluated three artificial neural networks topologies to improve the forecasting of photovoltaic power and energy. The results show that the topology with 5 neurons in the hidden layer was able to predict the photovoltaic energy with an error of less than 1.3% in relation to the measured energy.