A Review for the Development of ANN Based Solar Radiation Estimation Models

Abstract

Solar radiation estimation is the most vital part of solar system design. The solar system may be optimized if the radiation is estimated well in advance. Solar radiation is measured by devices such as Pyranometer, Pyrheliometer, Solarimeter, Radiometer, etc., installed at meteorological stations. Due to the unavailability of meteorological station at the locations of interest, solar radiation is estimated by means of estimation models. These models may be broadly classified as Mathematical/Statistical/Empirical models and the Soft Computing based models. These models accept meteorological variables such as wind speed, ambient temperature, relative humidity, cloud cover, etc., and geographical variables such as latitude, longitude, and altitude as input and provides Global Solar Radiation(GSR) at the output. Radiation estimation models are statistically tested and compared. The main aim of this paper is to briefly study and compare the Artificial Neural Network (ANN) based models. The paper deals with the basics of ANN along with its scope in solar radiation estimation. The study indicates that Artificial Neural Network (ANN) based models have significantly better accuracy than others. After the study of models, research gaps have also been pointed out in this paper to draw the attention of researchers.