Photo-Voltaic Panel Power Production Estimation with an Artificial Neural Network using Environmental and Electrical Measurements

Weather is one of the main problems in implementing forecasts for photovoltaic panel systems. Since it is the main generator of disturbances and interruptions in electrical energy. It is necessary to choose a reliable forecasting model for better energy use. A measurement prototype was constructed in this work, which collects in-situ voltage and current measurements and the environmental factors of radiation, temperature, and humidity. Subsequently, a correlation analysis of the variables and the implementation of artificial neural networks were performed to perform the system forecast. The best estimate was the one made with three variables (lighting, temperature, and humidity), obtaining an error of 0.255. These results show that it is possible to make a good estimate for a photovoltaic panel system.