Application Cases

Experiment Name: Detection of Blind Hole Damage in Aluminum Plates Using Scanning Laser Doppler Vibrometry and the YOLOv5s Deep Learning Model

Experimental Objective:
To address challenges in engineering structural damage detection such as complex signals, difficulties in imaging, and low efficiency of traditional analysis methods, this study proposes an intelligent damage detection method based on full guided wave field image target recognition. By integrating ultrasonic guided wave detection technology with deep learning algorithms, it systematically investigates the distortion characteristics of wave fields caused by damage and the underlying recognition mechanisms.

Test Equipment:
Scanning laser Doppler vibrometer, function generator, power amplifier ATA-2021H, piezoelectric transducer, reflective film, computer data processing system, and prefabricated aluminum plate specimens with blind hole damage.

Experimental Procedure:
This experiment established an SLDV (Scanning Laser Doppler Vibrometer) non-contact scanning platform to systematically validate the propagation characteristics of guided waves in damaged areas. It investigated wave field distortion patterns and damage recognition accuracy under multi-frequency excitation, analyzed the spatiotemporal evolution features of transient wave field images, and achieved precise localization and size identification of blind hole damage by training the YOLOv5s deep learning model.

Figure 1: Schematic Diagram of the Experimental Setup

Figure 2: Flowchart of the Experimental Setup

Experimental Results:
Under excitation with a 200kHz five-peak wave, the trained YOLOv5s model accurately captured wave field distortion features caused by damage, achieving sub-millimeter localization accuracy (error <1mm) and size recognition errors within 10%. Both numerical simulations and experimental validation demonstrated that the model effectively identified blind hole damage of different sizes, successfully avoided interference from excitation points, and exhibited strong generalization capabilities for damage types with wavenumber variation characteristics, such as metal corrosion and composite material delamination. By integrating deep learning with full wave field detection technology, this study provides a high-precision, high-reliability intelligent detection solution for engineering structural health monitoring.

Figure 3: Detection Results from Five Frames of Full-Field Wave Images in the Experiment

Figure 4: Comparison Between Experimental Transient Wave Field Detection Results and Actual Structural Blind Holes

Product Recommendation: ATA-2021B High-Voltage Amplifier

Figure: ATA-2021B High-Voltage Amplifier Specifications

This document is compiled and published by Aigtek. For more case studies and detailed product information, please continue to follow us. Xi'an Aigtek has become a large-scale instrument and equipment supplier in the industry with an extensive product line, and all demo units are available for free trial.