A GNN Based Voltage Sensitivity Perception Method for Distribution Networks with Variable Topology

Traditional data-driven voltage sensitivity analysis methods exhibit limited adaptability in response to network topology changes, which need to retrain the neural networks according to the updated topology information. To address this issue, this paper proposes a voltage sensitivity perception method based on graph neural networks (GNN). By embedding prior topology knowledge for information propagation, the method can effectively capture the feature mapping patterns of various topologies, thereby demonstrating improved model generalization ability in unknown topology scenarios. Matlab/Simulation results show that, compared to conventional data-driven algorithms, the proposed method achieves faster and more accurate sensitivity calculations in time-varying topology scenarios.