In order to improve the stability assessment of transient voltage of power system in case of topology variation and enhance the feature extraction ability in spatiotemporal aspects

a method combining graph convolutional network and recurrent neural network is proposed. Firstly

a graph convolutional network is introduced to represent power data

modeling the power system as a network structure and automatically learning the feature representations of voltage nodes. Then

the recurrent neural network is proposed to handle the temporal dependencies of transient voltage data and capture the temporal characteristics of transient voltage data. After that

an adaptive enhancement module is proposed to fuse the two output feature representations with each other and improve the spatiotemporal feature extraction ability of the model on the system topology structure. Finally

it is shown by numerical examples that the proposed method

compared to traditional assessment models

has higher prediction accuracy and effectiveness.