Power system risk assessment method based on hierarchical region graph neural network

By using a hierarchical regional graph neural network, the problems of the hierarchical relationship of power grid regions and the systemic risk propagation mechanism in the power system are solved, realizing the efficient fusion of multi-source data and multi-level risk assessment, and supporting the fine monitoring and decision-making of the power grid.

CN122242909APending Publication Date: 2026-06-19ZHEJIANG UNIV +1
1 Cites 0 Cited by

Patent Information

Application Number
CN202511953591.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-12-23
Publication Date
2026-06-19

AI Technical Summary

Technical Problem

Existing technologies are insufficient to effectively model the regional hierarchical relationships of power grids and the systemic risk propagation mechanism. They also have limited multi-source data fusion capabilities, lack multi-level risk assessment outputs, and are ill-suited to the new characteristics of the introduction of new energy power electronic equipment in new power systems and the volatility and randomness brought about by the high proportion of new energy access.

Method used

A hierarchical regional graph neural network is adopted. By dividing the power system into multiple regions, a local graph structure and inter-regional connection graph are constructed. The graph neural network and graph attention mechanism are used to learn and aggregate multi-source data to generate node-level, region-level and system-level risk assessment results.

Benefits of technology

It achieves efficient fusion of multi-source data and dynamic adaptation to topology changes, improves the model's robustness to new energy fluctuations and network structure changes, and provides multi-level risk assessment and accurate power grid risk early warning and scheduling decision support.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention relates to the technical field of electronic digital data processing, and discloses an electric power system risk assessment method based on a hierarchical region graph neural network, which comprises the following steps: dividing an electric power system into a plurality of regions, and constructing a local graph structure for each region; establishing a connection relationship between the regions to form a region-level graph; acquiring multi-source heterogeneous operation data of the power system as input characteristics of corresponding nodes; learning and aggregating node features in each local graph by adopting a graph neural network to obtain region feature representation of the corresponding operation region; aggregating the feature representation of each region in the inter-region connection graph based on a graph attention mechanism, and fusing risk information; and respectively generating node-level, region-level and system-level risk assessment results according to the feature representations of the nodes and the regions and the risk information. The problems of effective area layering, undefined risk propagation mechanism and lack of multi-level output are solved, and the purposes of multi-source data fusion, multi-level evaluation and dynamic adaptation to topological change are achieved.
Need to check novelty before this filing date? Find Prior Art