Solar Irradiance Forecasting using Decision Tree and Ensemble Models

Abstract

Sun's radiation is the pivotal driving force of the Earth and its prediction is quite significant for conducting numerous research projects in Renewable Energy Sources (RESs). The solar resource being an intermittent one, improvement in solar radiation prediction accuracy is strived for, to reduce uncertainty in RESs and enhance economical profits derived from them. This paper gives solar irradiance forecasting approach based on Decision Trees (DTs) and their ensemble models. Input data is comprised of 9 daily averaged meteorological parameters and 3 calendar variables for Chandigarh over 2 years (2017 & 2018). The implementation of forecasting models have been analyzed and compared based on Mean Square Error (MSE), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), Root Mean Square Error (RMSE), Correlation Coefficient (R-value). Pearson coefficient technique has also been used to assess the correlation between input features and solar irradiance. The model with least error metrics and highest R-value is considered to be optimal and is utilized to predict daily solar irradiance of Chandigarh for the year 2019.