Adaptive time-frequency analysis method for the analysis of dispersive elastic waves

Abstract

Although conventional methods such as short-time Fourier transform (STFT) and continuous wavelet transform (CWT) have been effectively used for analysis of dispersive elastic waves, rapidly varying wave signals may not be accurately analyzed by these methods. Because time-frequency tiling of conventional methods do not take into account dispersion phenomena, it is often difficult to trace accurately time-frequency varying feature of signals. The objective of this work is to develop advanced adaptive time-frequency analysis methods whose time-frequency tiling is varying to dispersion characteristics of signal to be analyzed. More specifically, a method called, the dispersion-based short-time Fourier transform (D-STFT) and dispersion-based continuous wavelet transform (DCWT) are newly developed. In these methods, each time-frequency tiling is adaptively rotated in time-frequency plane depending on estimated local dispersion rate of a wave signal. In proposed approach, dispersion relationship is estimated iteratively from ridge analysis of result by proposed adaptive methods where initial estimation is based on result by standard methods. To verify validity of present approach, Lamb waves measured in an aluminum plate were considered.