A neural network model of PV module temperature as a function of weather parameters prevailing in composite climate zone of India

Photovoltaic (PV) conversion of solar energy appears to be the most promising way of meeting the increasing energy requirements of the future, at a time when conventional sources of energy are being exhausted. Although the solar energy is an inexhaustible source of energy, the energy output of the PV module depends invariably on weather parameters like irradiance, relative humidity, wind speed, ambient temperature, etc. Understanding the dependence of PV output on these weather parameters will help us in setting up realistic goals for power production through non-conventional renewable sources. The most important observable, which affects the power output of a PV panel, is module temperature, which in turn depends on various weather parameters as listed above. Keeping this aim in mind, the present investigation focuses on developing a mathematical model using Artificial Neural Network (ANN) scheme to predict the module temperature using real-time field data of various parameters recorded at National Institute of Solar Energy (NISE), India, which is situated in composite climatic zone. The comparison of temperature obtained from the ANN model and the temperature recorded by the sensors deployed in the PV panels shows an overall coefficient of correlations ((Formula presented.)) of about 99.16% and a regression coefficient ((Formula presented.)) of 98.34%. These results indicate that ANN can be useful in generating good predictions based on completely unknown data consisting of several variables, which can be used to learn solar energy’s potential for any geographical location.