Abstract: Currently, the preventive control of power system transient stability is confronted with numerous challenges, including high model complexity and low solution efficiency. It is increasingly difficult to meet the actual needs of preventive control in terms of computational timeliness and model adaptability. Therefore, a new method of transient stability preventive control based on a graph convolutional network and a dominant action evaluation algorithm is proposed in this paper. Firstly, the graph convolutional network is employed to construct the transient stability evaluator, which can quickly and accurately evaluate transient stability by extracting features from power grid topology and operational data. Subsequently, according to the evaluation results, the dominant motion evaluation algorithm is used to learn and generate an optimal generator adjustment strategy. Finally, the effectiveness of the proposed method is validated through simulation experiments. Compared with existing methods, this method exhibits a higher level of adaptability to changes in power grid topology and provides new ideaes and technical approaches for the prevention and control of power system transient stability.