Application of Wavelet Analysis in Laser-generated Ultrasonic Nondestructive Testing

Laser-generated ultrasonic technology possesses incomparable advantages to many conventional testing methods of complete non-contact, fast response, non-destruction, wide band range, high precision and low cost, which is widely used to test the defects of aerocraft's components. However, the actual laser ultrasound signal is often mixed with a variety of noise and interference, which brings certain difficulty to material characterization and flaw identification. In order to de-noise laser-generated ultrasonic signal and enhance the detection sensitivity and reliability, wavelet analysis is applied to signal processing in this paper. The paper discusses the wavelet threshold de-noising methods, and the features of two traditional wavelet shrinkage methods----soft threshold and hard threshold, are analyzed in wavelet domain. Then an improved wavelet threshold function method is proposed on the basis of the two traditional threshold methods. According to the characteristics of the signal, db6 wavelet and 4 levels' decomposition are selected to complete the signal analysis. The functions of soft threshold, hard threshold and the improved threshold are employed in wavelet coefficients threshold processing respectively. Based on the analysis and comparison of each denoised figure and statistical property, the improved method can get a remarkable effect in signal de-noising is proved. It is concluded by theory analyzing and experiment research that the improved wavelet threshold function method has more advantages in laser ultrasonic signal de-noising compared with the two traditional threshold methods, which not only overcomes the shortcoming of discontinuity of hard threshold method, but also decreases the fixed bias between estimated wavelet coefficients and decomposed wavelet coefficients of soft threshold method.