A novel method for local frequency estimation of nonstationary random signals

A new time-frequency analysis approach based Hilbert transform and its application is analyzed. The comparison of two different time-frequency representations - wavelet and local spectral estimate - is to be established in this study. Conventional Fourier spectral analysis methods are insufficient for analyzing non-stationary data. The local frequency analysis is here proposed as an alternative. This paper illustrates that this method is adequate for non-stationery data and gives a more precise definition of particular events in time-frequency space than wavelet analysis. We can use the method to resolve changes in the frequency content of the data, as a function of time.