Solar Power Forecasting With Weather Factors Using Support Vector Regression

Abstract

Solar power forecasting is very important in ensuring correct and reliable operation of solar power grids. In forecasting the solar power output, it is not very easy to determine which parameters are crucial. In this work, the effect of using temperature at noon as well as cloud percentage at noon were used as inputs and solar power output was considered as the output. Effects of using different kernels were studied. It is noticed that when root mean square error (RMSE) and mean absolute error (MAE) values were considered linear and radial kernels produced relatively better results. Moreover having temperature data together with cloud percentage data produced more accurate modelling and enabled to determine more accurate solar power output with relatively lower RMSE and MAE values.