Application of Machine Learning Method for Selection of Cascade PLI Filter in Design of a Computerized ECG Signal Processing System

This article presents the results of the study of cascade PLI filters to justify the selection damping coefficient in order to increase the noise resistance system and create a software complex for ECG signal processing. To justify the selection of this coefficient, it is proposed to apply the supervised machine learning method. Theoretically significant analytical dependences processing system indicators on the selection damping coefficient of cascade PLI filters are obtained. It was found that at low values of this coefficient, the selectivity of narrow-band cascade PLI filters improves, and the quality indicators of ECG signal processing increase. Based on the analysis characteristics synthesized cascade PLI filter and the analysis of the processing results, the damping value of system
$\xi=0.1$
is justified and selected to create a software package. To confirm the practical significance of the improved processing approach, the performance software package, which combines the PLI filtration system, has been demonstrated. It is shown that the improved (developed) approach allows filtering natural power line interference with harmonics at frequencies of 50 Hz and 100 Hz and opens up new possibilities for ECG signal processing.