Improving distribution system state estimation: novel metadata driven approach for pseudo measurement generation

As the shift towards renewable energy progresses, the importance of accurately monitoring the distribution grid increases. Traditional monitoring has been challenged by the sparse deployment of measuring devices across medium and low voltage grids. This research introduces two novel AI-based approaches to generate realistic pseudo measurements (PMs) for medium-voltage busbars, utilizing consumer structure metadata and historical data. When these PMs are applied to estimate bus voltages in the medium-voltage grid, their accuracy proves comparable to that of historical measurements. However, in predicting line loadings, a forecasting error increase of up to 60 % is observed, though the prediction of critical line segments remains reliable using PMs alone. This study not only highlights the potential of PMs in enhancing grid monitoring but also suggests an optimized strategy for placing measuring devices to achieve the most accurate state estimation results.