Search Results

Abstract: The increasing use of solar photovoltaic (PV) systems in electrical power networks requires accurate solar irradiance forecasting for reliable and efficient power generation. This paper presents an efficient Random Forest-based Machine Learning technique for solar irradiance forecasting using historical irradiance and relevant meteorological parameters. The proposed model captures the nonlinear relationship between environmental conditions and solar irradiance and provides accurate prediction of available solar energy. The performance of the proposed technique is compared with a previous Deep Neural Network (DNN) method using accuracy and error rate as evaluation parameters. The Random Forest model achieves 99.67% accuracy with an error rate of 0.33%, while the previous DNN method achieves 94.70% accuracy with a 5.30% error rate. The results demonstrate that the proposed technique significantly reduces forecasting error and improves the reliability of solar irradiance prediction. Accurate forecasting can support PV power generation estimation, generation scheduling, power balancing, battery energy storage operation, voltage management, and renewable-energy integration. Overall, the proposed Random Forest technique provides an efficient and accurate solution for reducing uncertainty in solar PV generation and supporting reliable operation of modern electrical power systems.