Attention-Enhanced Deep Learning Model for Intelligent Solar Power Prediction and Sustainable Energy Management

AbstractThe growing demand for clean and sustainable energy has accelerated the adoption ofrenewable energy resources, particularly solar power. However, the intermittent nature ofsolar energy and its dependence on meteorological conditions create challenges for accuratepower prediction and efficient energy management. In this study, an attention-enhanced deeplearning framework is proposed for solar irradiance prediction using meteorological data. Theproposed model integrates Convolutional Neural Networks (CNN) for feature extraction,Long Short-Term Memory (LSTM) networks for capturing temporal dependencies, and anattention mechanism to emphasize important features influencing solar power generation.The model is trained and evaluated using meteorological data obtained from the NASAPrediction Of Worldwide Energy Resources (POWER) database, including parameters suchas solar irradiance, temperature, humidity, wind speed, and atmospheric pressure.Experimental results demonstrate that the proposed CNN–LSTM–Attention modeloutperforms baseline models including Artificial Neural Networks (ANN), CNN, andstandalone LSTM in terms of prediction accuracy and error metrics. The proposed frameworkprovides an effective approach for solar energy forecasting and can support intelligentrenewable energy management and sustainable energy planning.