Analyzing Multi-Area State Estimation in Power Systems in a Temporal Graph Convolutional Network Framework

The increasing availability of measurement data has resulted in the growing adoption of data-driven state estimation techniques in smart grids. However, the large volume of mea-surement data has raised concerns about data communication and processing, especially in smart grids with highly stochastic dynamics and stringent reliability requirements. To address these concerns, researchers are exploring new approaches in the field of distributed state estimation. This paper specifically focuses on investigating data-driven state estimation in smart grids using a temporal graph convolutional network framework within a multi-area setup. The performance of the distributed state estimation is evaluated by considering different modes of information sharing and graph structures in the multi-area model. A comparison is also made with the case of isolated localized models, where no information is shared among the regions. The presented analyses aim to provide insights into the role of the multi-area model in improving the performance of state estimation.