ANN Based Prediction of Module Temperature in a Single Axis PV System

Abstract

Photovoltaic technology is one of the most effective and cleanest methods of obtaining energy from the sun. The efficiency of modules that is one of the most basic components of photovoltaic systems, is very sensitive to environmental variables. The air temperature and the electric current that passing through the panels cause the panels to heat up. Temperature negatively affects the efficiency of the panels. In photovoltaic systems, the panel temperature must be determined in order to know the rate of heat losses, which is one of the many types of losses. In this study, an artificial neural network model was developed that determines the temperature of a single axis tracking solar panel according to environmental variables. For the development of the model, 36699 data rows were used, which were measured and recorded with a data logger for one year. In this model, the module temperature is the dependent variable while the solar irradiance, ambient temperature, wind speed, relative humidity and panel power are selected as the independent variables. In the developed model, there is a 98.87% correlation between the actual values and the estimated values. The developed model predicts the module temperature very well according to the actual values with 1.45 MAE, 4.27 MAE, 6.37% MAPE and 2.24% RSE performance criteria. By knowing the module temperature, the amount of heat losses that will occur in photovoltaic systems can be calculated. In addition, estimating the panel temperature value can be used as an important parameter in the organization of cooling processes to increase efficiency.