Browsing by Author "Lovato Caiza, Washington Gregorio"

Now showing 1 - 1 of 1
  • No Thumbnail Available
    Item
    ” implementación de un sistema fotovoltaico residencial OFF-GRID con predicción energética”
    (Ecuador: Latacunga: Universidad Técnica de Cotopaxi (UTC), 2025-07-29) Herrera Robayo, David Ramiro; Lovato Caiza, Washington Gregorio; Castillo Fiallos, Jessica Nataly
    This study presents the implementation of an OFF-GRID residential photovoltaic system for a type D house located in Chimborazo Province, Ecuador, where there is no access to the conventional power grid. Two predictive models were used to estimate solar radiation: second-degree polynomial regression and a decision tree, using hourly data from the NASA POWER portal covering the years 2019 to 2024. The decision tree model showed better performance in terms of accuracy, with a Mean Squared Error (MSE) of 0.212 and a Coefficient of Determination (R²) of 0.8561, compared to the polynomial model which achieved an MSE of 0.6359 and R² of 0.6599. The prediction estimated a maximum solar radiation of 416.21 W/m² at 12:00 PM in January. These results are crucial for designing efficient solar energy systems in rural areas without grid access. The research demonstrates that machine learning models offer significant advantages for predicting complex phenomena such as solar radiation, enabling better planning for renewable energy use and improving the quality of life in isolated communities