• Sangita Solanke
• Dr. Tushar T. Phartade

Accurate solar power forecasting is essential for optimizing energy production from photovoltaic (PV) systems and ensuring efficient integration of solar energy into the grid. However, predicting solar energy output remains a challenge due to the intermittent and non-linear nature of solar radiation. This paper presents a hybrid ensemble approach combining multiple neural network architectures and ensemble techniques—bagging, boosting, and stacking—to enhance forecasting accuracy. By leveraging the strengths of individual models, this hybrid approach captures complex patterns in data, accounting for various factors like weather patterns, seasonal variations, and solar radiation. The methodology involves several stages: data collection and preprocessing, feature selection, model development, ensemble formation, and model evaluation. We utilize historical meteorological data, apply advanced neural network models such as Multi-Layer Perceptron (MLP), Recurrent Neural Networks (RNN), and Long Short-Term Memory (LSTM), and use ensemble techniques to improve model accuracy. The model’s performance is evaluated through metrics like Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and R².Results show that the hybrid ensemble model outperforms traditional methods, with an accuracy of 90.32%, precision of 98.95%, recall of 91.08%, and an F1 score of 94.85%. The analysis of feature importance reveals that key meteorological variables, such as shortwave radiation and wind speed, significantly influence solar power predictions. Despite the strong performance, there is room for improvement in reducing false negatives, especially in predicting low solar power values. This approach offers valuable insights for enhancing solar power forecasting, contributing to the efficient integration of renewable energy into the grid.

"HYBRID APPROACH FOR THE ENSEMBLES OF NEURAL NETWORKS FOR SOLAR POWER FORECASTING", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.12, Issue 4, page no.m305-m313, April-2025, Available :http://www.jetir.org/papers/JETIR2504C40.pdf

"HYBRID APPROACH FOR THE ENSEMBLES OF NEURAL NETWORKS FOR SOLAR POWER FORECASTING", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.12, Issue 4, page no. ppm305-m313, April-2025, Available at : http://www.jetir.org/papers/JETIR2504C40.pdf