A GraphSAGE Model for Spectral Feature Analysis of Wideband Oscillations^*

Wideband oscillations in power systems span a broad frequency range and pose significant challenges to grid stability. Traditional spectral analysis methods, such as Fourier Transform and Wavelet Transform, have limitations in handling non-stationary signals and meeting real-time requirements. To address these challenges, this paper proposes a GraphSAGE-based graph neural network model for wideband oscillation identification. By categorizing oscillation data and training the model, GraphSAGE effectively extracts oscillation features and enables dynamic monitoring.Experimental results demonstrate that the GraphSAGE model outperforms traditional methods in accuracy, computational efficiency, and adaptability, providing a reliable and efficient solution for rapid wideband oscillation detection.