A Study on the Applicability of LSTM Networks for Smart Grid Monitoring and Dynamic Security Analysis

To address the challenges in real-time monitoring and dynamic security assessment of modern power systems, this study proposes an intelligent solution centered on the Long Short-Term Memory (LSTM) network. This solution establishes two models: one for precise short-term load forecasting and another for rapid transient stability assessment.Experimental results show that in short-term load forecasting, the model’s Mean Absolute Percentage Error (MAPE) was reduced by approximately $24 \%$ compared to the best-performing baseline model (Extreme Gradient Boosting (XGBoost)). In transient stability assessment, the model achieved an accuracy of $98.3 \%$ and reduced assessment time to the millisecond level, an efficiency improvement of over three orders of magnitude compared to traditional methods. Compared to the advanced Transformer model, this study’s LSTM model achieved comparable accuracy with approximately ${2 0 - 2 5 \%}$ fewer parameters and less training overhead, demonstrating higher efficiency and engineering applicability.