Abstract: Deep learning (DL) models have been proven to be highly effective in detecting false data injection attacks (FDIAs) in smart grids relying only on the measurements of the secured component.

Abstract: This paper presents an retrieval-augmented based question-answering model tailored for power grid dispatch regulations. We introduce an automated high-quality fune-tuning dataset generation ...