Simulated CIGRE MV datasets for physics-informed distribution system state estimation

AC power-flow solutions and simulated measurement views for distribution system state estimation on the balanced 15-bus CIGRE medium-voltage benchmark with photovoltaic and wind generation. The operating scenarios cover 26 August to 21 October 2019 at 15-minute resolution (5,376 timestamps) and four renewable penetration levels (0, 20, 40 and 60%), giving 21,504 converged AC cases. The measurement views include a nominal view, an MLP baseline view and 32 comparison views that vary renewable penetration, sensor coverage, packet loss, sensor noise and pseudomeasurement uncertainty. Each view provides sensor placement, clean and corrupted measurements with their standard deviations and availability masks, and fixed chronological splits. These datasets support the article "Physics-Informed Graph Learning for Distribution System State Estimation: Accuracy, Physical Consistency, and Training Efficiency under Renewable Variability and Limited Observability", submitted to Advances in Applied Energy. They were generated with the code at https://github.com/velas02/physics-informed-dsse. The source series from the ENTSO-E Transparency Platform (TransnetBW control area, compiled by Open Power System Data) are third-party data: they are not redistributed here and are not covered by this licence. See README.md for attribution and for how to rebuild the profiles.