Solar radiation forecast using hybrid SARIMA and ANN model

Abstract

Solar Energy have an enormous potential for generating renewable electricity. In the tropics solar energy are abundance all year long but suffer from uncertainty caused by rain and clouds. Accurate prediction of solar radiation can increase the affectivity and productivity of solar energy sources. Monthly average of solar radiation data are obtained from stations in Malaysia. The data are modeled using the Seasonal Autoregressive Integrated Moving Average (SARIMA) model, artificial neural network (ANN) model and Hybrid ANN and SARIMA model. The SARIMA model is a reliable tool in forecasting seasonal data, on the other hand the ANN model have been proven to be a good model in forecasting non-linear data. By combining both model a more accurate model can be obtained. Finally the forecasting performance each model is compared by using mean absolute error (MAE), the mean absolute percentage error (MAPE) and root mean square error (RMSE). The result shows that the hybrid model is better in forecasting solar radiation data.