Factorization of processes parametric spectra on the base of multiplicative linear prediction polymodels

The linear prediction models can be useful in different tasks of statistical radio engineering. The examples of multimode spectra decomposition on separate components for speech signals, heart rhythmograms, reflected ultrasound signals and hydroacoustic signals have been shown in the paper. The calculation of autoregressive coefficients and parametric power spectrum density of the multiplicative models were derived. Factorization of spectrum estimations is shown using an example of a multiplicative linear prediction model. Using the models that have been developing in our research it is possible to develop the new methods of complex processes analysis. The methods of rhythmogram analysis can be useful for specialists, who create algorithms of cardiogram analysis. More specifically, we consider a method of multimode spectrum factorization in composite process on components using our multiplicative linear prediction polymodels.