Mode identification of broadband Lamb wave signal with squeezed wavelet transform

Multiple wave modes often exist in the ultrasonic guided waves simultaneously, and these modes are dispersive, so the guided wave signals are very complex, even for the relatively simple situation of a narrowband excitation. The guided wave signals are even more difficult to analyze for broadband excitations. Time-frequency representations are appropriate for the analysis of the guided wave signals considering their non-stationary and transient nature. As a post-processing tool, the squeezed wavelet transform is studied for broadband Lamb wave mode identification in this work. The influence of the parameters of the Gabor mother wavelet on the performance of the transform is analyzed in details. It is found that the product of the σ parameter of the used Gauss function and the center frequency ω0 of the wavelet decides the overall time and frequency resolutions, so a proper selection of the value of this product σω0 is crucial for the squeezed wavelet transform. The squeezed wavelet transform is first applied to the analysis of a synthesized signal for verification. Then it's applied for mode identification of a simulated broadband Lamb wave signal. By traversing the value of σω0, a roughly optimum analysis performance is achieved for the squeezed wavelet transform for the case of σω0=11, where the modes are well separated and the interferences between the modes are minimal. It's proved that as an alternative tool, the squeezed wavelet transform could be used for the analysis of a broadband Lamb wave signal. An additional benefit of this transform is that it permits reconstruction of the original signal or its components, which is not possible for the reassigned scalogram.