Time–Frequency Analysis of Echoes Signal in Ultrasonic Testing of Adhesion Based on Short-Time Fourier Transformation

While assess the bonding quality of thin plate, it is quite difficult to extract the characteristics of echoes for analysis because of the weakening of defect echoes which were interfered by the high reflection of metal layer and noise. Target on the non-stationary and time variant characteristic of echoes, this paper propose to use the short time Fourier transformation methods to conduct these kinds of assessment. First, this paper constructs a echoes mathematical model and simulate different kinds of defect echoes for different levels of debonding. These simulation data provide the theoretical base for the quantitative assessment of bonding quality. Base on the time intervals of the echoes, this paper divide the superposed echoes into different parts and calculate the short-time Fourier spectrogram then extract the characteristics of spectrogram. The results indicate that there are distinctive differences on echo spectrogram for different levels of bonding quality. The characteristics which were extracted from the sample signals can be used to assess the bonding quality quantitatively.