Prediction of power produced by solar panels with artificial neural networks by using environmental factors

The output power of a photovoltaic panel is dependent on environmental factors such as solar irradiance, air temperature, wind speed, wind direction, relative humidity etc. This dependence is nonlinear and is due to the differences in the production techniques of photovoltaics. This necessitates detailed experimental studies for the detection of photovoltaic panel characteristics. Robust prediction of the power generated by photovoltaic panels is crucial in the proper planning of power generation systems. In this study, it was tried to estimate the power value produced by photovoltaic panels using artificial neural networks. In order to make a more robust prediction in experimental work, the power values produced by photovoltaic panels were measured and recorded for one year, considering the environmental variables such as the amount of solar radiation, air temperature, wind speed, wind direction, relative humidity and solar elevation angle. In the study, fixed and single axis tracking monocrystalline panels were used. The predicted values obtained by the developed artificial neural network models were compared with the measured values and the obtained findings were examined. When the developed artificial neural network models were tested with data that were not used in the network training process, it was observed that very robust predictions were performed with the Root Mean Squared Error (RMSE) error rates not exceeding 1.4% for the fixed panel and 2.34% for the single axis tracking panel. In the developed artificial neural network models, it is observed that the correlation coefficient between the dependent power variable and the environmental variables was very high with a value between 99.637% and 99.998%. Furthermore, in this study, the developed models were tested by Multiple Linear Regression which is one of the traditional methods. The results showed that the models obtained by using environmental variables predicted the power value produced by the photovoltaic panels with high accuracy. This study was carried out in Batman province; however, at any location in the world, the energy production of any planned photovoltaic installations can be estimated with high accuracy using environmental variables obtained from meteorological stations.