Data Augmentation for Dynamic Security Assessment based on Hybrid Model-Data Driven Approach

In response to the increasing complexity and scale of power systems, this paper proposes a dynamic security assessment (DSA) framework based on a hybrid model-data-driven approach. This framework aims to enhance the efficiency and accuracy of DSA by incorporating the flexibility of data-driven methods. The paper focuses on transient stability assessment, a crucial task within DSA, which refers to a power system’s ability to return to its initial stable operating state or transition to a new stable operating state after experiencing a significant disturbance. To more accurately generate critical samples and improve the efficiency and accuracy of neural network training, a novel data augmentation method is introduced in this paper. Additionally, critical intervals are constructed, and the interpretability of neural networks is enhanced based on the physical meaning of transient stability assessment.