Voltage Sensitivity Estimation in Distribution Networks with Incomplete Measurements and Unknown Topologies

Given the challenges of unknown topology and the lack of voltage measurement devices at certain nodes in rural distribution networks, this paper proposes a data-driven voltage sensitivity fitting method combining Generative Adversarial Networks (GAN) and the Least Squares Method (LSM). First, GAN is used to generate voltage data for unmeasured nodes based on voltage measurements from partial photovoltaic (PV) nodes and network-wide load data. Subsequently, LSM is employed to fit a voltage sensitivity matrix without relying on topology information. Simulation analysis based on a standard distribution network case demonstrates that the proposed method can accurately fit the voltage sensitivity matrix under conditions of partial voltage measurement data loss and unknown network topology.