A novel approach for short-term prediction of solar and wind power production using recurrent neural networks and empirical mode decomposition

The rising global energy demand and environmental challenges posed by fossil fuels have intensified the need for efficient renewable energy solutions. To address this, researchers are leveraging Artificial Intelligence (AI) and Machine Learning (ML) to enhance energy forecasting and optimization. However, existing models often face limitations in accuracy and computational efficiency. This study proposes a novel hybrid approach combining Empirical Mode Decomposition (EMD) with advanced ML techniques—including Gated Recurrent Unit (GRU), Long Short-Term Memory (LSTM), and their bidirectional variants (BiLSTM and BiGRU)—to improve solar and wind power predictions. Meteorological data, such as solar radiation and wind speed components, were decomposed using EMD to extract meaningful patterns. The modified portfolio method was then applied to integrate forecasts from different ML models, optimizing overall accuracy. The results demonstrated significant improvements, achieving 87% accuracy in wind speed prediction and 80% in solar radiation forecasting, outperforming conventional methods. These advancements highlight the potential of AI-driven models in enhancing renewable energy integration and grid stability. By increasing prediction reliability, this research supports the transition toward sustainable energy systems, offering a viable solution to global energy and environmental challenges.