TCSET'2010, February 23-27, 2010, Lviv-Slavske, Ukraine 280 Digital Filters for Power Spectral Density Estimation of Heart Rhythm Volodymyr Falendysh, Bohdan Yavorskyy, Mykhaylo Bachynskyy Abstract - In this paper digital filters for a real-time estimation of the power spectral density of stationary components of a heart rhythm are considered. Keywords – Heart rhythm, monitoring system, human functional state, power spectral density, stationary component. I. INTRODUCTION Increasing of complexity of human-machine automation systems in different branches of living activity necessitated the problem of control and evaluation of reliability of a human operator as one of the most important element of such systems [1]. Nowadays the estimation of power spectral density of heart rhythm is considered as one of the most informative and accurate methods of evaluation the human functional state [2]. The hearth rhythm is a variable, nonstationary sequence of values of electrocardiosignal RR-intervals, particularly with a periodically stationarity. Under suppose about a relatively long with specific conditions of a working time of an operator the filter method for analysis of the power spectral density of rhythmosignals previously was considered [3]. In this paper the main concept of digital filters development for an automatic real-time monitoring system of the human functional state is presented. II. PARAMETRIC SPECTRUM OF HEART RHYTHM The main feature of the heart rhythm monitoring system is that heart rhythm is non-stationary and is considered as periodically correlated stochastic process (PCSP) under breath systems influence of a working man. This determines the structure of the monitoring system. Periodically correlated discrete stochastic process represent via stationary components ) ( ȟ k n , like in [4]: ｦ  ｸ ｸ ｹ ｷ ｨ ｨ ｩ ｧ  Z k n N ik n n K ʌ 2 exp ) ( ȟ ) ( ȟ k (1) where K N - period of correlation of sequence. So, when hearth rhythm sequence is PCSP, than its stationary component ) ( ȟ k n is a special sample with capture of selection from ) ( ȟ n through K N samples. Heart rhythm sequence ) ( ȟ n is represented in frequency domain via power spectral density (PSD) of stationary components ) ( ȟ k n - parametric spectrum ) , Ȧ ( S n that is periodic with period K N . Parametric spectrum ) , Ȧ ( S n estimations are obtained using filter banks [3, 5]. Each filter bank consists of K N digital filters, but instead of Chebyshev pass-band filters [3], digital filters with the structure shown on Fig.1 are considered. Fig.1 Structure scheme of digital filter used for PSD estimation. Each filter in the bank has resonant frequency i Ȧ and bandwidth const ǻȦ  . Transmission function of the filter is like in [6]: 2 2 1 1 1 1 2 1 ) (        Z b Z b Z a a Z H (2) Coefficients b 1 and b 2 of each digital filter are determined using [6]. Coefficients a 1 and a 2 (let 2 1 a a   ) are determined by normalization requirement : 1 ) Ȧ (  i j H . III. CONCLUSION Application of the filter method for estimation of PSD of the heart rhythm sequence allows us for provide of its a long time analysis with low mean square errors in the time interval. Considered structure of digital filter provide us better effectiveness comparatively to Chebyshev pass-band filters. REFERENCES [1] C.B. Greeves Human Factors in Action. The FLYLEAF, Summer 2002, p. 24-26. [2] J. Fahrenberg, C.J.E. Wientjes. Recording methods in applied environments. In: R.W. Bachs, W. Boucsein (Eds.) Engineering Psychology: Issues and Applications. London: Lawrence Erlbaum Associates, 2000, p. 111-136. [3] Ɇ.ȼ. Ȼɚɱɢɧɫɶɤɢɣ, ɘ.Ɂ. Ʌɟɳɢɲɢɧ, ȼ.ȼ. Ɏɚɥɟɧɞɢɲ. Ɏɿɥɶɬɪɨɜɢɣ ɦɟɬɨɞ ɜɢɡɧɚɱɟɧɧɹ ɩɚɪɚɦɟɬɪɿɜ ɜɚɪɿɚɛɟɥɶ- ɧɨɫɬɿ ɫɟɪɰɟɜɨʀ ɪɢɬɦɿɤɢ. // ȼɿɫɧɢɤ ɯɦɟɥɶɧɢɰɶɤɨɝɨ ɧɚɰɿɨɧɚɥɶɧɨɝɨ ɭɧɿɜɟɪɫɢɬɟɬɭ, 2006, ɬ. 1. ɋ 180-185. [4] ə. Ⱦɪɚɝɚɧ. ȿɧɟɪɝɟɬɢɱɧɚ ɬɟɨɪɿɹ ɥɿɧɿɣɧɢɯ ɦɨɞɟɥɟɣ ɫɬɨɯɚɫɬɢɱɧɢɯ ɫɢɝɧɚɥɿɜ. Ʌɶɜɿɜ, ɐɟɧɬɪ ɫɬɪɚɬɟɝɿɱɧɢɯ ɞɨɫɥɿɞɠɟɧɶ ɟɤɨ-ɛɿɨ-ɬɟɯɧɿɱɧɢɯ ɫɢɫɬɟɦ., 1997, 361ɫ. [5] ɘ.ɂ. Ƚɪɢɛɚɧɨɜ, ȼ.Ʌ. Ɇɚɥɶɤɨɜ. ɋɩɟɤɬɪɚɥɶɧɵɣ ɚɧɚɥɢɡ ɫɥɭɱɚɣɧɵɯ ɩɪɨɰɟɫɫɨɜ, Ɇ., «ɗɧɟɪɝɢɹ», 1974, 240 ɫ. [6] Ȼ.ɂ. əɜɨɪɫɤɢɣ, Ɂ.ɂ. Ⱦɨɦɛɪɨɜɫɤɢɣ. Ɋɚɫɱɟɬ ɰɢɮɪɨɜɵɯ ɩɨɥɨɫɨɜɵɯ ɮɢɥɶɬɪɨɜ ɬɢɩɚ ɑɟɛɵɲɟɜɚ // Ɋɚɞɢɨɬɟɯɧɢɤɚ, 1981ɝ. ɬ. 36 ʋ10, ɋ.79-81. Volodymyr Falendysh, Bohdan Yavorskyy, Mykhaylo Bachynskyy – Ternopil Ivan Puluj State Technical University, Ruska Str., 56, Ternopil, 46001, UKRAINE, E-mail:falendysh@gmail.com