Dynamic State Estimation for Distribution Networks Considering Large-Scale Missing Measurement Data Reconstruction

To address widespread and prolonged measurement data gaps in distribution networks, a dynamic state estimation method for distribution networks considering large-scale missing measurement reconstruction is proposed. In the offline training stage, a measurement reconstruction model is built and trained based on Wasserstein generative adversarial network (WGAN), which learns the correlation between measurements and the fluctuation law of loads, so as to accurately reconstruct randomly or continuously missing measurements. In the online training stage, the replenished measurement data sets are fed into the extended kalman filter (EKF) state estimator to realize the reliable state estimation of the distribution network under insufficient measurements. Case studies demonstrate that the proposed method enhances measurement redundancy and network observability while maintaining computational efficiency, which is feasible for engineering applications.