An Ensemble Supervised Machine Learning Model for Solar Irradiance Prediction Using Tree-Based Learners

To facilitate renewable energy to be more deeply integrated into the current power system’s controls, an accurate solar energy prediction is crucial. Data-driven algorithms offer a chance to enhance solar power forecasts because of the increased granularity of data available. In this paper, the solar irradiance prediction has been done with the ensemble stacked learning method where three machine learning models, decision tree, extreme learning, and random forest are considered as the base learners and the linear regression has been used as the final estimator for the Jharkhand state. The model performance has been evaluated with various statistical errors and it has been analyzed that the stacked model shows the best accuracy of 96.64% and root mean squared error of 50.94.