The problem of enhancing the effectiveness of spectral analysis of signals observed against the background of noise by the Root-MUSIC method has been considered using the surrogate data technology implemented by adapting the algorithm of phase randomization of the Fourier transform samples of the initial data in relation to the signal-to-noise ratio (SNR). The proposed variant of the surrogate data technology was shown to be effective at low values of SNR and a small number of samples. At large values of SNR an additive variant of the surrogate data technology actually does not cause an emergence of surrogate interference typical for nonadaptive variant. The proposed variant of the technology can be applied in combination with other methods of spectral analysis.
Theproblem of processing correction of signals observed against the background of noise has been considered in relation to their spectral analysis by the Root-MUSIC method using the technology of surrogate autocovariance functions (ACF) of observation. The results of simulation modeling are presented dealing with the correction of observation processing by means of the phase randomization of spectral components of observation ACF using the ATS-algorithm for generating the observation surrogate ACF. It has been shown that in generating surrogate data from the observation ACF, ATS-algorithm ensures the highest efficiency at small signal-to-noise ratios.