Power spectral analysis of signal with time-varying characteristics

Abstract EN

During the past two decades, the spectrogram method has widely used to analyze the characteristic of non-stationary signals.

In this paper, two of the power intensity of the signal algorithms for time-frequency characteristic of non-stationary signal are proposed, tested and compared to standard schemes.

The first algorithm is based on the continuous time-frequency characteristic and the second is based on the discrete time-frequency characteristic.

The performance of each algorithm is evaluated and compared to the existing standard spectrogram analysis.

The results show that the proposed power intensity waveform analysis provides very good performance in both continuous and discrete time - frequency characteristic for classical evaluation criteria, and for the preservation of data statistics.

The Fourier transform of the intensity of signals is computed in continuos and discrete form.

Two methods for the digital implementation of the proposed algorithm are suggested and described.

Moreover, It is shown that this algorithm is useful for detecting possible nonstationarity of signals.

Comparisons of the result with multiresolution analysis reveals that the prepared algorithm has better frequency resolution where as the wavelet transform shows higher time resolution.