Solar Irradiance Forecasting Using Ensemble Models of Machine Learning

Abstract

Ensemble learning is widely used in forecasting due to its unique ability to combine individual models to improve prediction power. It provides the composite prediction where the final accuracy is better than the accuracy of the individual models. This paper will discuss different ensemble learning models for solar irradiance forecasting. Nine daily average meteorological parameters are considered a feature input of Jaipur over three years for solar irradiance forecasting. Based on different parameters like Mean Square Error (MSE), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), Root Mean Square Error (RMSE), and Correlation Coefficient (R2), comparative analysis of different ensemble learning models are made. The Principal Component Analysis and Feature Importance plot methodology have been used for input feature selection. The model with the lowest error metrics and greatest R-value is regarded as the best and is used to forecast Jaipur's daily solar irradiance for 2021.