A Self Supervised State Calculation of Distribution Network Based on Mask Graph

As the demand for electricity keeps rising, distribution networks are expanding in scale and becoming more complex. Accurately and efficiently assessing the state of these networks is crucial for ensuring the safe, stable, and economical operation of power systems. Conventional methods for calculating the state of distribution networks typically rely on physical models of the power system. These approaches involve extensive mathematical computations and simulation analyses, making it challenging to meet the real-time requirements of modern power systems. To tackle this problem, this paper presents an innovative approach for calculating the state of distribution networks, which is grounded in heterogeneous graph mask modeling. The proposed method employs a heterogeneous graph neural network to model the distribution network, mapping its nodes and edges to corresponding elements within the graph structure. It then uses graph mask technology to reconstruct node attributes and edges through self-supervised mechanisms. Leveraging a pre-trained graph neural network model, the state of the distribution network can be computed rapidly and precisely. The experimental results demonstrate that this method can effectively capture the topological relationships and state variables of distribution networks, achieving high accuracy in downstream tasks. In contrast to traditional methods, it exhibits greater adaptability and scalability, making it a more suitable solution for contemporary power system applications.