Ya Zar Doe - Post-graduate Student, Department of Biomedical Engineering, Southwest State University (Kursk). E-mail: SFilist@gmail.com O.V. Shatalova - Ph. D. (Eng.), Associate Professor, Department of Biomedical Engineering, Southwest State University (Kursk). E-mail: shatolg@mail.ru L.V. Pleskanos - Ph. D. (Eng.), Associate Professor, Southwest State University (Kursk). E-mail: SFilist@gmail.com

At singular decomposition of an electrocardiosignal from it allocate four additive components: trend, «fast waves», «slow waves» and noise. At certain cardiovascular pathologies the noise component of singular decomposition represents a certain interest and it demands a choice of the way allowing to integrate its properties into space of signs of the classifying model. The way of formation of the space of informative signs intended for neural network qualifiers of cardiovascular pathologies, Fourier consisting in calculation of window transformation of structural functions of a noise component of singular decomposition of an elec-trocardiosignal is developed. The way differs in that structural functions are divided into frames, each frame corresponds to a relevant harmonica (a slow wave) which is present at a signal. In a frame the set of structural functions is defined, each of which differs in a step of sampling of structural function which varies within 20 dB. Defining Fourier\'s range of the corresponding structural function of a frame, allocate significant frequencies which collect in a frame with the corresponding weight coefficients from structural to structural function. The vector of informative signs received thus is intended for the corresponding block of the neural network qualifier of cardiovascular pathology.

 

1. SHatalova O.V., ZHilin V.V., Zar Do JAa. Sposob vydelenija medlennykh von iz shumovojj sostavljajushhejj singuljarnogo razlozhenija ehlektrokardiosignala // Nauchnyjj vzgljad na sovremennyjj ehtap razvitija obshhestvennykh, tekhnicheskikh, gumanitarnykh i estestvennykh nauk. Aktualnye problemy: sbornik statejj po itogam Vseros. nauchno-praktich. konf. SPb.: Izd-vo «KultInformPress». 2014. S. 137−141.
2. Volkov I.I., Emeljanov S.G., Filist S.A. Metod klassifikacii slozhnykh obektov na osnove analiza strukturnykh funkcijj medlennykh voln // Biomedicinskaja radioehlektronika. 2012. № 4. S. 6−11.
3. Filist S.A., Tomakova R.A., Gorbatenko S.A.i dr. Strukturno-funkcionalnye reshenija nechetkikh nejjronnykh setejj dlja intellektualnykh sistem analiza raznotipnykh priznakov // Fundamentalnye i prikladnye problemy tekhniki i tekhnologii. 2011. № 1(285). S. 85−92.
4. D-Urso P., Giordani P. A least squares approach to principal component analysis for interval valued data // Chemometrics and Intelligent Laboratory Systems. 2004. V. 70. P. 179−192.
5. Giordani P., Kiers H.A.L. Principal Component Analysis of symmetric fuzzy data // Computational Statistics and Data Analysis. 2004. V. 45. P. 519−548.