Unmanned cluster cascade failure deduction model and method based on physical information graph

By using a physical information graph-based cascading fault simulation model for unmanned clusters, combining physical information with graph neural networks, the problem of rapidly and accurately simulating cascading faults in unmanned cluster fault diagnosis is solved, achieving real-time updates of cluster status and stability and robustness in fault prediction.

CN122154495BActive Publication Date: 2026-07-21HEFEI UNIV OF TECH
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
HEFEI UNIV OF TECH
Filing Date
2026-05-09
Publication Date
2026-07-21

AI Technical Summary

Technical Problem

Existing unmanned cluster fault diagnosis technologies struggle to achieve efficient fault prediction while ensuring physical consistency, especially in rapidly and accurately simulating the propagation of cascading faults in large-scale clusters.

Method used

A fault inference model for unmanned clusters based on physical information graphs is adopted. By integrating physical information with graph neural networks, fault inference is performed using the cluster adjacency matrix and state matrix. The model is trained by combining multiple loss functions to achieve real-time updates of cluster state and accurate judgment of fault truth.

Benefits of technology

It enables rapid and accurate simulation of unmanned cluster cascade faults, improves the stability and robustness of fault prediction, and can update the cluster status in real time in dynamic environments, reducing the instability and false alarm rate of fault propagation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122154495B_ABST
    Figure CN122154495B_ABST
Patent Text Reader

Abstract

The present application relates to unmanned cluster fault diagnosis and health management technical field, especially a kind of unmanned cluster cascade fault deduction model and method based on physical information graph.The present application first obtains fault deduction model, and fault deduction model is based on the cluster adjacency matrix and cluster state matrix of current time k and deduces the cluster state matrix of next time k+1;On time k+1, the cluster state matrix is updated in cycle by fault deduction model, and the state true value vector of cluster is derived according to cluster state matrix, and state true value vector is used to describe whether each node is fault;Until the state true value vector on adjacent two cycles on time k+1 is consistent, then the latest cluster state matrix is output as prediction result.The present application fuses physical information and graph neural network to carry out cluster fault deduction, realizes the rapid, accurate simulation to cluster cascade fault.
Need to check novelty before this filing date? Find Prior Art