Solar Radiation Forecasting Using Support Vector Regression

Solar energy is the most predominant renewable energy resource available to humankind. To remain depend on it in future, forecasting of solar energy is essential. In this paper, solar potential is forecasted with the help of Support vector regression (SVR) depending on other easily measurable parameters. The parameters like pressure, temperature, humidity are exploited in the prediction of daily global solar radiation. The data used for the study is taken for a period of two year for the location of New Alipore, Kolkata. Two models where developed using RBF kernel and Polynomial kernel function of SVR. The performance of this two models are evaluated with the statistical measures viz, Coefficient of Determination (R-2) and Root Mean Square Error (RMSE). The result obtained are R-2 of 0.7976 and RMSE of 1.0564 for training while R-2 of 0.7845 and RMSE of 1.0532 for testing with RBF kernel. While polynomial kernel gives R-2 of 0.9393 and RMSE of 1.1975 for training while R-2 of 0.9060 and RMSE of 1.1594 for testing.