Power System Transient Stability Analysis Based on Transformer and Supervised Contrastive Learning

Transient stability analysis (TSA) is crucial for maintaining the stability of power systems. Compared to traditional dynamic simulations, neural network-based power system TSA models have been widely applied in recent years due to their strong nonlinear mapping capability and fast prediction speed. In this paper, a new TSA approach is proposed based on the Transformer neural network, which combines the encoder-decoder architecture and the attention mechanism. A multi-channel feature extraction structure and a supervised contrastive learning algorithm are utilized to mitigate the overfitting phenomenon and enhance the generalization capability of the proposed model, resulting in competitive performance. The efficacy of the proposed TSA method is validated by the superior performance in the standard IEEE 39-bus test case.