Sensitivity analysis of 9 models for estimating the power of photovoltaic monocrystal and polycrystalline panels

As the energy generation with photovoltaic systems is exponentially growing, the need for prediction of generation power is much more important than ever. In this paper, based on the experimental data of the solar site along one-year, the photovoltaic systems of monocrystalline and polycrystalline panels for 2.5 kW solar power plant at Vali-e-Asr University of Rafsanjan, has been modelled with artificial neural network. To obtain the optimal model for the desired panels, different variables have been considered as the input and output of the model, while all variables are: panel temperature, direct radiation, the output power and the kind of panels (monocrystalline and polycrystalline). Also sensitivity analysis has been performed for different input and output parameters as well as for the number of different layers of neurons and functions. The results show that the model with the input of temperature and radiation and the output of the power of monocrystalline and polycrystalline panel is the most accurate model with lowest error.