Multi-Step forecasting of renewable energy short-circuit ratio in new power systems using LSTM with temporal attention mechanism

Abstract With the rapid integration of renewable energy sources (RES) into modern power grids, maintaining system stability and reliability has become increasingly challenging. The Renewable Energy Short Circuit Ratio (RESCR) serves as a critical metric for assessing system strength under fluctuating RES conditions. Traditional methods, such as static short circuit ratio calculations, fail to capture dynamic changes in system strength. This study introduces a novel multi-step forecasting model combining Long Short-Term Memory (LSTM) networks with an attention mechanism to predict RESCR values accurately. The model captures long-term dependencies while providing interpretable insights through dynamic attention weights. Experimental results on real-world datasets demonstrate that the proposed model significantly outperforms standard LSTM models, particularly in multi-step prediction tasks. By enhancing prediction accuracy and interpretability, this study contributes to improving grid stability and maximizing renewable energy utilization.