On zero-shot learning in neural state estimation of power distribution systems
This work addresses a domain-specific problem for power distribution systems, but it is incremental as it builds on existing graph neural network methods with data augmentations and hyperparameter tuning.
The paper tackles neural state estimation in power distribution systems by addressing the inability of models to adapt to grid changes like sensor loss and branch switching in a zero-shot fashion, proposing data augmentations and conducting a grid search that confirms graph neural networks' robustness but shows deeper networks do not always perform better.
This paper addresses the challenge of neural state estimation in power distribution systems. We identified a research gap in the current state of the art, which lies in the inability of models to adapt to changes in the power grid, such as loss of sensors and branch switching, in a zero-shot fashion. Based on the literature, we identified graph neural networks as the most promising class of models for this use case. Our experiments confirm their robustness to some grid changes and also show that a deeper network does not always perform better. We propose data augmentations to improve performance and conduct a comprehensive grid search of different model configurations for common zero-shot learning scenarios.