1Ph.D. Research Scholar, LNCT University, Bhopal, India. Email: nehaamitkhare@gmail.com
2Professor, LNCT University, Bhopal, India. Email: anuragrai@lnct.ac.in
The use of renewable energy in Indian Telecom sites is increased in recent times for enhancing the energy efficiency of networks. This study proposes a machine learning–based framework for accurately predicting the power output of a simulated hybrid wind and solar photovoltaic (PV) energy system. The methodology integrates advanced regression techniques, including feature selection, boosting methods, Gaussian process regression, and stacked ensemble learning, to enhance prediction performance. A comparative analysis is conducted between the hybrid wind–PV system and a standalone PV configuration to evaluate the benefits of hybridization. Synthetic datasets are generated to model the power output of the hybrid system by incorporating wind-related parameters, such as wind speed and torque measurements, alongside PV-related variables, including solar irradiance, solar load, and ambient temperature. Data preprocessing involves normalization and feature selection using the ReliefF algorithm to identify the most influential input variables. Multiple machine learning models are trained and evaluated. The best-performing model for the hybrid system is identified as a stacked ensemble with 0.97017 R² Values. This model stacked predictions from boosted trees and Gaussian process regression. The hybrid system's performance is compared to a PV-only model, revealing superior prediction accuracy for the hybrid configuration.
Key Words: Hybrid Solar-Wind System, Telecom Networks, Machine Learning, Power Prediction, Wind Torque, Solar load, Boosted Trees, Gaussian Process Regression, Stacked Ensemble.
How to cite this article: Khare N, Rai ASD. Hybrid Solar - Wind Power Prediction Using Advance Stacked Ensemble Learning: System For Indian Telecom Networks. Int J Drug Deliv Technol. 2026;16(8s): 880-895; DOI: 10.25258/ijddt.16.8s.97
Source of support: Nil.
Conflict of interest: None