Related Experiment Video
Updated: Jun 22, 2025
08:08
Evaluating Targeting Accuracy in the Focal Plane for an Ultrasound-guided High-intensity Focused Ultrasound Phased-array System
Published on: March 6, 2019
5.2K
Utilizing improved YOLOv8 based on SPD-BRSA-AFPN for ultrasonic phased array non-destructive testing.
1State Key Laboratory of Mechanical System and Vibration, Shanghai Jiao Tong University, China.
Ultrasonics
|June 29, 2024
Summary
This study enhances defect detection in phased array ultrasonic testing (PAUT) using an improved YOLOv8 deep learning model. The new algorithm significantly boosts accuracy for challenging defects like side-drilled holes, advancing automated non-destructive testing (NDT).
Area of Science:
- Materials Science and Engineering
- Non-Destructive Testing (NDT)
- Artificial Intelligence in Engineering
Background:
- Phased array ultrasonic testing (PAUT) is crucial for industrial non-destructive testing (NDT), but manual defect analysis is inefficient and unreliable.
- Current automated data collection in PAUT contrasts with manual data interpretation, highlighting a need for advanced computer-based solutions.
- Deep learning object detection shows promise but requires extensive annotated data, which is scarce in NDT, limiting detection of low-resolution defects.
Purpose of the Study:
- To improve the accuracy and efficiency of defect detection in phased array ultrasonic testing (PAUT) using deep learning.
- To address challenges in detecting low-resolution images and defects with varying sizes and positions.
- To develop a robust and computationally efficient algorithm for automated ultrasonic image analysis.
Main Methods:
- Proposed an enhanced YOLOv8 algorithm incorporating space-to-depth convolution (SPD-Conv) to preserve information in low-resolution images.
- Integrated a bi-level routing and spatial attention module (BRSA) into the backbone for richer multiscale feature representation.
- Replaced the original neck structure with an asymptotic feature pyramid network (AFPN) to reduce model complexity and parameters.
Main Results:
- The enhanced algorithm achieved superior detection of flat bottom holes (FBH) and aluminum blocks on simulated datasets compared to baseline YOLOv8.
- Demonstrated significant improvements for side-drilled holes (SDH), achieving an F1 score of 82.50% (17.56% increase) and IOU of 65.96% (0.43% increase).
- On experimental datasets, FBH detection reached an F1 score of 75.68% (9.01% increase) and IOU of 83.79%, with robust performance against noise and high computational efficiency.
Conclusions:
- The proposed intelligent defect detection algorithm significantly enhances the accuracy and efficiency of PAUT image analysis.
- The integration of SPD-Conv, BRSA, and AFPN effectively addresses limitations in detecting low-resolution and complex defects.
- This advancement contributes to the development of smart industry applications through reliable automated NDT solutions.
Keywords:
Convolutional neural networksDefect detectionNon-destructive testingPhased array ultrasoundYOLOv8More Related Videos
Related Concept Videos
01:17NMR Spectrometers: Radiofrequency Pulses and Pulse Sequences
790
A pulse is a short burst of radio waves distributed over a range of frequencies that simultaneously excites all the nuclei in the sample. Upon passing a radio frequency pulse along the x-axis, the nuclei absorb energy corresponding to their Larmor frequencies and achieve resonance. This shifts the net magnetization vector from the z-axis toward the transverse plane. This angle of rotation of the magnetization vector, or the flip angle, is proportional to the duration and intensity of the pulse.
790
01:10Reconstruction of Signal using Interpolation
191
Signal processing techniques are essential for accurately converting continuous signals to digital formats and vice versa. When a continuous signal is sampled with a period T, the resulting sampled signal exhibits replicas of the original spectrum in the frequency domain, spaced at intervals equal to the sampling frequency. To handle this sampled signal, a zero-order hold method can be applied, which creates a piecewise constant signal by retaining each sample's value until the next...
191