Related Experiment Video
Updated: Sep 11, 2025
08:51
Statistical Modelling of Cortical Connectivity Using Non-invasive Electroencephalograms
Published on: November 1, 2019
5.7K
Bridge Damage Identification Using Time-Varying Filtering-Based Empirical Mode Decomposition and Pre-Trained
Shenghuan Zeng1, Jian Cui2, Ding Luo1
1Shenzhen Expressway Engineering Testing Co., Ltd., Shenzhen 518000, China.
Sensors (Basel, Switzerland)
|August 14, 2025
Summary
This study introduces a novel framework for bridge damage identification using time-varying filtering empirical mode decomposition (TVFEMD) and convolutional neural networks (CNNs). The method improves signal quality and enhances damage classification accuracy for better bridge health monitoring.
Area of Science:
- Structural Engineering
- Signal Processing
- Artificial Intelligence
Background:
- Bridge health monitoring is crucial for operational safety and maintenance.
- Current methods face challenges with poor signal quality, feature extraction, and classification accuracy.
Purpose of the Study:
- To develop an advanced framework for accurate bridge damage identification.
- To overcome limitations in signal processing and classification for bridge health monitoring.
Main Methods:
- Integration of time-varying filtering-based empirical mode decomposition (TVFEMD) with pre-trained convolutional neural networks (CNNs).
- Adaptive denoising and time-frequency reconstruction to enhance signal features and suppress noise.
- Comparative analysis of different CNN models (ResNet-50) for damage classification.
Main Results:
- TVFEMD demonstrated superior frequency separation and modal purity compared to traditional EMD.
- ResNet-50 achieved optimal performance in damage classification with TVFEMD-processed signals.
- Principal Component Analysis (PCA) visualization confirmed improved feature clustering and separability.
Conclusions:
- The proposed TVFEMD-CNN framework significantly enhances bridge damage identification accuracy.
- TVFEMD effectively preprocesses signals, improving CNN model adaptability and recognition.
- This approach offers a robust solution for practical bridge health monitoring challenges.
