Version 1
: Received: 29 July 2024 / Approved: 30 July 2024 / Online: 30 July 2024 (13:47:03 CEST)
How to cite:
Ramírez-Rivera, F. A.; Guerrero-Rodríguez, N. F. Ensemble Learning Algorithms for Solar Radiation Prediction in Santo Domingo: Measurements and Evaluation. Preprints2024, 2024072439. https://doi.org/10.20944/preprints202407.2439.v1
Ramírez-Rivera, F. A.; Guerrero-Rodríguez, N. F. Ensemble Learning Algorithms for Solar Radiation Prediction in Santo Domingo: Measurements and Evaluation. Preprints 2024, 2024072439. https://doi.org/10.20944/preprints202407.2439.v1
Ramírez-Rivera, F. A.; Guerrero-Rodríguez, N. F. Ensemble Learning Algorithms for Solar Radiation Prediction in Santo Domingo: Measurements and Evaluation. Preprints2024, 2024072439. https://doi.org/10.20944/preprints202407.2439.v1
APA Style
Ramírez-Rivera, F. A., & Guerrero-Rodríguez, N. F. (2024). Ensemble Learning Algorithms for Solar Radiation Prediction in Santo Domingo: Measurements and Evaluation. Preprints. https://doi.org/10.20944/preprints202407.2439.v1
Chicago/Turabian Style
Ramírez-Rivera, F. A. and Néstor F. Guerrero-Rodríguez. 2024 "Ensemble Learning Algorithms for Solar Radiation Prediction in Santo Domingo: Measurements and Evaluation" Preprints. https://doi.org/10.20944/preprints202407.2439.v1
Abstract
Solar radiation corresponds to fundamental parameters for solar photovoltaic (PV) technology. Reliable solar radiation prediction became valuable to design solar PV systems, performance, operatively efficient planning, safety operation, grid dispatch and financial characteristics. However, high quality ground-based solar radiation measurements are scarce, especially for very short-term time horizon. Most of the existent studies trained the machine learning (ML) model used dataset with time horizon of 1-hour or day, very fewer studies have been reported using a dataset with a 1-mitute time horizon. In this study, a comprehensive evaluation of nine ensemble learning algorithms (ELA) is performed to estimate the solar radiation in Santo Domingo with a 1-minute time horizon dataset collected from a local weather station. The ensemble learning evaluated included seven homogeneous ensembles; Random Forest (RF), Extra Tree (ET), Adaptive Gradient Boosting (AGB), Gradient Boosting (GB), Extreme Gradient Boosting (XGB), Light Gradient Boosting (LGBM), Histogram-based Gradient Boosting (HGB) and heterogeneous ensemble named as Voting and Stacking. RF, ET, GB, HGB were combined to develop Voting and Stacking and Linear Regression (LR) was adopted in the second layer of the Stacking. Five technical metrics, Mean Squared Error (MSE), Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE) and Coefficient of Determination (R2) were used as criteria to determine the prediction quality of the developed ensemble algorithms. Comparison of the results indicates that the HGB algorithm offers superior prediction performance among the homogeneous ensemble learning and overall, the Stacking provides the best accuracy with metric values of MSE=3218.27, RMSE= 56.73, MAE=29.87, MAPE=10.60, R2=0.964.
Keywords
ensemble learning; evaluation metrics; heterogeneous ensemble learning; homogeneous ensemble learning; hyperparameter; time horizon; solar radiation
Subject
Engineering, Energy and Fuel Technology
Copyright:
This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.