Practical Dynamic Security Region Model: A Hybrid Physical Model-Driven and Data-Driven Approach

Security analysis methods based on security region (SR) play a pivotal role in power system analysis. The construction of security margin contributes towards the mining of geometric and physical characteristics of a practical dynamic SR (PDSR). The data-driven approach provides strong support for power system security analysis. However, poor interpretability and weak generalization ability of artificial intelligence method have hindered its application in power systems. Although the model-driven method can effectively analyze the internal mechanisms of physical models, improve their interpretability, and enhance the reliability of security margin assessment, its strong nonlinear structure and low computational efficiency limits its further application. This study first constructed a reliable security margin based on the characteristics of the SR, and consequently obtained the critical operating interval based on a reliability indicator. Finally, data-driven method was used to modify the security margin of PDSR. The proposed method was validated on the New England 10-machine-39-node system and a practical system in China. The results confirmed its higher evaluation accuracy compared to existing methods. Further, it could effectively explain the model mechanism, providing key factors that affect the evaluation results. A PDSR with high accuracy and interpretability will facilitate the accurate formulation of transient stability assessment results, while increasing the response time for power system dispatch decision-making.