A Sub/Super-Synchronous Oscillation Localization Method Based on Graph Attention Network with Physical Feature Embedding

With the continuous increase in the capacity of grid-connected new energy sources such as wind and photovoltaic power, the issues of sub-synchronous oscillation and super-synchronous oscillation caused by the interaction between power electronic devices and the grid have become increasingly prominent. Therefore, accurately localizing online oscillation sources is of great importance for preventing the expansion of accidents. In this paper, a modal parameter identification method is first proposed. By selecting the real part of the synchrophasor as the characteristic quantity, the precise decoupling and identification of Sub/Super-SO modal parameters are realized. On this basis, a physical feature-embedded graph attention network localization method is proposed, in which the high-precision modal parameters obtained from identification are embedded as physical features into graph nodes, and the attention mechanism is used to adaptively learn the oscillation propagation patterns in the grid topology. Finally, simulation verification based on the IEEE 14-bus system demonstrates that the proposed method can effectively achieve accurate localization of oscillation sources.