A Frequency Stability Prediction Method Using Spatial Temporal Graph Convolutional Networks and Self-attention Mechanism

Frequency stability prediction (FSP) in power systems is crucial for formulating effective post-fault control measures promptly and accurately. However, the traditional data-driven methods fail to effectively incorporate the system spatial-temporal characteristics into the model training, and suffers from insufficient utilization of system information, poor generalization ability in the face of new topology and interpretability. To address these above problems, a FSP prediction method based on self-attention mechanism and Spatial-Temporal Graph Convolutional Network (STGCN) is proposed in this paper. Firstly, the proposed modified STGCN prediction method utilizes graph convolutional layer to capture the topology of the system to obtain spatial information. Then, the time series data of the system is captured by using a one-dimensional convolutional layer to obtain the temporal information. Finally, a differentiable self-attention graph pooling layer based on self-attention mechanism is used to enhance the generalization ability and robustness of the modified STGCN model. Test results on a modified New England 39-bus system validate that the effectiveness of the proposed modified STGCN.