Effective Defect Features Extraction for Laser Ultrasonic Signal Processing by Using Time–Frequency Analysis

Abstract

The time-frequency analysis (TFA) by wavelet transform is adopted for the laser ultrasonic signal processing, and the effective features extraction of the material defect is obtained. The TFA is adopted here to analyze the laser-generated surface acoustic wave (SAW) signal which contains the defect features, the echo wave features are extracted significantly, especially under the condition of low signal-to-noise rate (SNR). The simulation model by using finite element method (FEM) is set up in an aluminum plate with different surface defect depths in detail, and the defect depths prediction with TFA is also considered. It shows that, without extra denoising process, the echo SAW is extracted significantly in case of defect depths ranging from 0.1mm to 0.9mm at SNR of -3dB by TFA. The TFA for processing the laser ultrasonic signal provides a promising way to get the defect information, with the accuracy increased by 7.9dB in this work, which is extremely meaningful for the ultrasonic signal processing and material evaluation.

Keywords

• Time-frequency analysis
• laser ultrasonic signal
• feature extraction
• wavelet transform
• signal-to-noise rate