Power system transient stability assessment based on hierarchical graph pooling method considering missing data

• The initial filling of missing values based on k-order neighborhood mean. • High-precision data reconstruction using a skip-connected masked graph autoencoder. • Differentiated processing based on the node structure and electrical properties. • Extracting multi-scale subgraph features for transient stability assessment (TSA). • The accuracy of TSA is maintained in the case of various types of data missing. The transient stability assessment (TSA) model based on graph deep learning relies on complete system topology and its characteristics. However, electrical operation data may be missing during the measurement and transmission process, which can lead to a decline in model evaluation performance. To address this issue, this paper proposes a hierarchical graph pooling TSA method for power systems that accounts for missing data. First, the power system is modeled as a graph with topological connections, and the missing data is filled using the K-order neighborhood mean (KNM). Next, a masked graph autoencoder with jumping knowledge is developed to reconstruct the missing features. Finally, considering both the topological attributes of nodes and the temporal characteristics of electrical quantities, a well-designed graph pooling method is introduced. During hierarchical graph pooling, subgraph features of the power grid at different scales are extracted and fused to achieve accurate and reliable TSA. Test results on the IEEE 39-bus system and a provincial power system in China demonstrate that the proposed method can maintain high evaluation performance under various types of missing data, exhibiting strong robustness and practicality.