Time-varying spectral analysis of non-stationary signals based on combined wavelet and Fourier transforms
Mustafa M. Matalgah, Jerome Knopp · International Journal of Electronics · 1995
The wavelet transform (WT) has received attention in recent years in the analysis of non-stationary signals. Studies have shown that the WT provides very good resolution in the time direction and comparatively poor resolution in the frequency direction for broadband signals. On the other hand, the Fourier transform (FT) provides better resolution in the frequency direction, but it is not suitable for the analysis of transients. A new algorithm is presented to analyse non-stationary signals efficiently. It is based on combined wavelet and Fourier transforms. This algorithm makes use of the WT to display the time domain signal components and the FT to display their spectral components. Further a new method is suggested to eliminate the interference terms between the different scales in wavelet analysis. The algorithm is applied to computer simulation and real data examples. Results suggest that it provides comparable temporal and frequency resolution of transients.