To quickly and accurately identify the dominant instability mode of a power system

this paper proposes a power system dominant instability mode identification method based on a Graph Convolutional Network (GCN) and Bi-directional Gated Recurrent Unit (Bi-GRU). Firstly

according to the temporal evolution law and spatial distribution characteristics of transient electrical quantities before and after the system fault

the characteristic matrix representing the operation state of the power system is constructed. Then

a deep learning model combining GCN and Bi-GRU is established. GCN is used to integrate topological spatial information to improve the model's generalization. At the same time

Bi-GRU is used to adaptively perceive the global time series information of input features to deeply mine the spatial and temporal characteristics of the feature matrix. In this way

the model clarifies the deep connection and interaction among transient electrical quantities in the transient process

and realizes the accurate identification of the dominant instability mode of the power system. Finally

the experimental results of the modified IEEE-39 node system and the actual power grid in a certain area show that the proposed method has a certain degree of interpretability and offers advantages in effectiveness

accuracy

and adaptability compared with other deep learning methods.