Non-contact structure strain intelligent measurement method based on dynamic visual energization

By combining event cameras and long short-term memory neural networks, the problems of ambient light interference and insufficient accuracy of optical methods for structural strain measurement in dynamic environments are solved, realizing non-contact strain monitoring with high temporal resolution and providing a high-precision real-time health monitoring solution for engineering structures.

CN121739913APending Publication Date: 2026-03-27XI AN JIAOTONG UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-12-17
Publication Date
2026-03-27

AI Technical Summary

Technical Problem

Existing technologies for structural strain measurement in dynamic environments suffer from limitations such as optical methods being susceptible to ambient light interference, high equipment costs, and existing event camera methods failing to effectively capture the time dependence of strain evolution, resulting in insufficient measurement accuracy.

Method used

Event flow data during the strain process of a structural surface is acquired using an event camera. A four-element spatiotemporal feature dataset is constructed by filtering ROI regions, time sliding window segmentation, and spatiotemporal feature extraction. An end-to-end strain mapping mechanism is established using a long short-term memory neural network to achieve high temporal resolution non-contact strain monitoring.

Benefits of technology

It achieves highly robust and high-precision non-contact structural strain measurement in dynamic environments, breaking through the bottleneck of traditional optical measurement and providing reliable technical support for real-time health monitoring of engineering structures.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121739913A_ABST
    Figure CN121739913A_ABST
Patent Text Reader

Abstract

A non-contact intelligent structural strain measurement method based on dynamic visual energization belongs to the technical field of structural strain measurement, and comprises the following steps: firstly, synchronously acquiring structural surface event stream data and a strain label; a focus deformation sensitive area is screened through the ROI; dividing the event stream into sub-event streams matched with the strain sampling frequency by using a time sliding window; extracting quaternary spatio-temporal characteristics (event rate, positive and negative polarity proportion, spatial gradient and event interval kurtosis) with clear mechanical significance; constructing a three-order time sequence feature tensor input two-layer LSTM network; a SmoothL1 loss function is adopted to optimize model parameters; precise mapping from historical visual features to strain values is realized; according to the invention, an end-to-end event flow-strain mapping mechanism is constructed, and the surface strain of a to-be-measured structure is continuously monitored by using a dynamic visual sensor (event camera) in a dynamic measurement environment.
Need to check novelty before this filing date? Find Prior Art