Related Experiment Video
Updated: Jul 12, 2026
08:27
Image Recognition and Parameter Analysis of Concrete Vibration State Based on Support Vector Machine
Published on: January 5, 2024
Hybrid Ultrasonic Framework for CFRP-Steel Interfacial Defect Classification via Wavelet Packet Transform and
Zhuo Huang1, Zhuoyi Chen1, Sheng Tan1
1School of Civil and Environment Engineering, Changsha University of Science and Technology, Changsha 410114, China.
Ultrasonics
|July 10, 2026
Summary
This study enhances defect detection in carbon fiber reinforced polymer (CFRP) steel structures using phased array ultrasonic testing (PAUT) and artificial neural networks (ANNs). A novel approach achieved 95.83% accuracy in identifying interfacial bonding defects.
Area of Science:
- Civil Engineering
- Materials Science
- Non-Destructive Testing
Background:
- Carbon Fiber Reinforced Polymer (CFRP) composites strengthen steel structures but are susceptible to interfacial bonding defects.
- Conventional ultrasonic testing faces challenges with noise and limited feature extraction for reliable defect identification.
Purpose of the Study:
- To develop an intelligent framework for classifying interfacial defects in CFRP-steel hybrid structures.
- To improve the reliability and accuracy of defect detection beyond conventional ultrasonic methods.
Main Methods:
- Integration of phased array ultrasonic testing (PAUT) with encoder-assisted acquisition.
- Application of Wavelet Packet Transform (WPT) for signal denoising and feature extraction.
- Training and comparison of three Backpropagation Artificial Neural Network (BP-ANN) variants for defect classification.
Main Results:
- Eight statistical energy-based features were extracted using WPT, enhancing signal discriminability.
- The BP-ANN optimized with the conjugate gradient algorithm achieved a 95.83% overall accuracy in defect classification.
- Encoder-assisted acquisition improved the consistency of PAUT signal collection.
Conclusions:
- The proposed WPT-based feature engineering and BP-ANN framework offers a feasible approach for intelligent interfacial defect classification.
- The study demonstrates the effectiveness of advanced signal processing and machine learning in enhancing non-destructive testing for composite structures.
- Results provide a reference for future research on manufacturing defects and practical engineering applications.
