Artificial neural network applied to estimate the power output of bipv systems
Registro en:
10.9790/0661-1901017378
instname:Universidad de Bogotá Jorge Tadeo Lozano
reponame:Repositorio Institucional de la Universidad de Bogotá Jorge Tadeo Lozano
Autor
Aristizábal, A.J.
Institución
Resumen
This paper presents an artificial neural network (ANN) model to estimate the power generated by integrated photovoltaic systems in buildings - BIPVS. The model has as primordial variables, the solar radiation and the ambient temperature of the site of installation of the photovoltaic generator and integrates secondary variables such as the zenith solar angle and the azimuth solar angle. The artificial neural network consists of three layers of operation that allows to adapt to the behavior of the environmental and electrical variables of the photovoltaic generator to create output variables of electrical power through daily profiles.
The neural network was implemented in the software Matlbab™ and it was validated using the actual data of monitoring of a 6 kW BIPV system installed at Universidad de Bogotá Jorge Tadeo Lozano, in Bogotá, Colombia. The results indicate a correlation coefficient of 98% on the output power of the BIPV system between the artificial neural network and the performance data of the solar photovoltaic plant. These results show the reliability of the model for PV systems operating in different climatic conditions and different generation capacities.