"

Modeling And Prediction Of A Multivariate Photovoltaic System, Using The Multiparametric Regression Model With Shrinkage Regularization And Extreme Gradient Boosting.

Published in: Prospective and trends in technology and skills for sustainable social development. Leveraging emerging technologies to construct the future: Proceedings of the 19th LACCEI International Multi-Conference for Engineering, Education and Technology
Date of Conference: July 19-23, 2021
Location of Conference: Virtual
Authors: Saul Huaquipaco Encinas (Universidad Nacional del Altiplano, PE)
Jose Cruz (Universidad Nacional del Altiplano, PE)
Norman Jesus Beltran Castañon (Universidad Nacional de Juliaca, PE)
Ferdinand Pineda (Universidad Nacional del Altiplano, PE)
Christian Romero (Universidad Nacional del Altiplano, PE)
Julio Fredy Chura Acero (Universidad Nacional del Altiplano, PE)
Wilson Mamani Machaca (Universidad Nacional del Altiplano, PE)
Full Paper: #557

Abstract:

Alternative energy systems have more frequently been acquiring a fundamental role in the generation of energy that promotes the development of countries in social, economic, and environmental terms. For the efficient operation of photovoltaic systems (SFV), it is necessary to make predictions about their operation, turning them into intelligent systems. The present work proposes the collection, modeling, and prediction of a multivariate SFV, using a multiparametric regression model, presenting five regression models with machine learning: three that use Shrinkage regularization and two that use eXtreme Gradient Boosting (XGBoost). Results obtained, we note that the five predictions have determination coefficients higher than 99.47%; being XGBoost with n_estimators = 500 which reduces the root mean square error by about 55%. Likewise, in all cases, the test times are less than 1 second. The results were validated so that they not only have mathematical significance, but are also real, showing that XGBoost with n_estimators = 10 does not meet the five validation conditions, so this prediction model should not be considered.