In the context of increasingly complex load consumption and generation structures in low voltage (LV) networks, driven by the adoption of new energy use cases (e.g., heat pumps, electric vehicles, photovoltaic), accurate household load estimation is crucial for grid observability and effective congestion management. This work addresses the challenge of estimating highly stochastic household loads in LV grids, particularly under scenarios with incomplete smart meter coverage (in the range from 10% to 90%), reflecting real-world conditions.

Building on previous research that developed a clustering framework for household load profiles using both consumption data and additional household information (e.g., number of residents), this study presents two key contributions:

Improved Estimation Method: We refine the existing clustering-based estimation approach, enhancing its accuracy and reliability in diverse coverage conditions.

Integration of Grid Information: A novel method is introduced that combines feeder measurements with representative load profiles derived from household clusters, demonstrating better performance in load estimation.

The introduced methodology leverages publicly available data, including partially field-based datasets, and applies a benchmark network model tailored to German LV grid structures. The findings underline the inherent difficulty of precise load estimation due to high demand variability and highlight the necessity of hybrid approaches—integrating both clustering techniques and grid-level measurements—to achieve robust estimations in various conditions.

This research contributes to the advancement of grid management strategies, providing valuable insights for enhancing the observability and operational efficiency of LV networks (e.g., grid congestion management) amidst evolving energy landscapes.