Implementación de métodos de predicción de radiación solar para una zona particular de la geografía colombiana

Abstract

In the present work, two prediction techniques implemented to predict values ​​of incident solar radiation in the municipality of El Paso, located in the department of Cesar – Colombia. The selected techniques have a different nature among them. By On the one hand, there is the traditional statistical analysis technique and on the other, the implementation of a model based on Machine Learning. The ARIMA model (acronym for the English autoregressive integrated moving average, in Spanish, autoregressive integrated moving average model mobile) is chosen as representative of the statistical aspect because it is the most complete due to which combines the autoregressive (AR), integrated (I) and moving average (MA) components; Taking into account the database of the study area, an ARMA model (acronym from English autoregressive moving average model, in Spanish, model of moving average autoregressive) because there was no need to apply the built-in component for the data treatment. The implemented model of Machine Learning for the prediction of solar radiation is that of Neural Networks, due to the great virtues it has being This is a suitable tool for solving problems of this nature.