Methods, equipment and systems for locating bridge damage areas

By using a parallel physical information neural network structure with macroscopic and microscopic dual channels, combined with fully connected neural networks and graph neural networks, the problem of environmental interference masking minor damage in bridge monitoring was solved, and early accurate location and uncertainty quantification of bridge damage areas were achieved.

CN121787284BActive Publication Date: 2026-05-26贵州宏信创达工程检测咨询有限公司
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
贵州宏信创达工程检测咨询有限公司
Filing Date
2026-03-04
Publication Date
2026-05-26

AI Technical Summary

Technical Problem

Existing bridge monitoring technologies cannot accurately identify early minor damage, and there is a problem that damage sensitivity and environmental interference are difficult to decouple, resulting in the inability to locate the damaged area in a timely manner.

Method used

A parallel physical information neural network structure with macroscopic and microscopic dual channels is adopted. The global predicted physical field quantities are generated through a fully connected neural network, and the thermoelastic mechanical control equation and stress-strain constitutive relation are introduced for physical constraints. Combined with graph neural network and Bayesian neural network, the bridge damage area is located.

Benefits of technology

It enables early and accurate localization of bridge damage areas, decouples the masking effect of environmental changes on damage, and improves the accuracy and confidence of damage identification.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121787284B_ABST
    Figure CN121787284B_ABST
Patent Text Reader

Abstract

This application relates to the field of bridge safety monitoring technology, specifically to a method, device, and system for locating bridge damage areas. The method monitors bridge health using a parallel macro-micro dual-channel physical information neural network structure. The macro-channel generates global predicted physical field quantities based on global monitoring information through a fully connected neural network. The micro-channel, based on the global predicted physical field quantities output by the macro-channel, introduces a damage mechanics constitutive model to physically constrain the graph neural network, focusing on key component areas, identifying local physical inconsistencies, thereby locating local damage and quantifying uncertainty. This achieves decoupling from the environment / load-dominated response provided by the macro-channel, solving the problem in existing technologies where environmental changes mask changes caused by minor damage.
Need to check novelty before this filing date? Find Prior Art