Approximation of Empirical and ANN Based Solar Radiation Models for Four Smart Cities in Tamil Nadu with Most Prompting Input Parameters

Abstract

The abundant source of Solar Radiation is possible in different latitude of Tamil Nadu. The several Empirical models have developed and accuracy of the model is further validated with Artificial Neural Network models. The ANN and Empirical model is developed with different combination of input attributes. The Approximate Sunshine Solar Radiation, Temperature and Relative Humidity are the input attributes. In this paper, it has been evaluated that combination of Approximate Sunshine Solar radiation and Temperature ANN based model is best for prediction of Daily Global Solar Radiation when compared to ST based Empirical model. The Correlation analysis is performedfor selection of most relevant input attributes. And hence the estimation of daily global solar radiation is performed with minimum number of input attributes. The training and testing of each model is performed by ANN Simulink model. And from this trainingofinputvariables,theoverfittingofthemodelisreduced. The estimated GSR is validated with experimental data which is collected from Tamil Nadu AgricultureUniversity.