У дисертації розв'язано актуальну наукову задачу обґрунтування вибору математичної моделі та розроблення методів опрацювання біосигналів, які дають можливість виділення інформативних ознак намагання реалізувати пацієнтами комунікативну функцію в структурі електроенцефалографічних та електроміографічних сигналів для задачі компенсації порушеної комунікативної функції людини.Обґрунтовано вибір математичної моделі таких біосигналів у вигляді кусково стаціонарного випадкового процесу. Розроблено методи опрацювання таких сигналів в межах трансляцій ковзного вікна, для виявлення часових моментів початку та закінчення процесу мовлення а також виявлення ознак наявності основного тону в структурі цих сигналів під час намагання реалізації порушеної комунікативної функції. Обґрунтовано вибір інформативних ознак початку та закінчення процесу мовлення та наявності основного тону.^UIn the dissertation the actual scientific problem of substantiation of the choice of mathematical model and development of methods for processing biosignals, which allow to select informative features of individual voice phonemes in the structure of electroencephalographic and electromyographic signals to compensate for impaired human communication function, is solved.Based on the biophysical foundations of the communicative function, a method of its compensation in cases of its violations is proposed. The method is based on the selection and processing of electroencephalographic signals taken from the surface of the patient's head near the speech centers of the brain, and electromyographic signals taken from the surface of the neck near the vocal folds.A new application of a piecewise stationary random process as a mathematical model of electroencephalographic and electromyographic signals is substantiated, which is adequate to the physical nature of such signals and has the means to highlight informative features that are indicators of the process of communicative function. Based on a well-founded mathematical model, methods for statistical processing of electroencephalographic and electromyographic signals within a sliding window are developed, which are based on calculation and subsequent estimation of power spectral density distributions within sliding window translations and allow to detect time moments signals) and identify signs of the presence of the main tone (based on the results of processing electromyographic signals) during the implementation of human communicative function. It is established that the obtained values of variation of the average estimates of the power spectral density distribution within the sliding window are informative signs of the beginning and end of the speech process, and the average estimates of the electromyographic signal power spectral density distribution in the range of the fundamental frequency are calculated. signs of the presence of the main tone.The structure of the system of selection of electroencephalographic and electromyographic signals is offered. Such signals were selected using the Neurocom electroencephalographic complex as well as the electromyographic signal selection unit, which is a single channel for amplifying the biopotentials of the Neurocom encephalograph with extended bandwidth. Due to the significant noise level in the structure of electromyographic signals, their filtration was performed using the method of spectral subtraction.The results of further processing of the selected biosignals confirmed the correctness of the choice of mathematical model of electroencephalographic and electromyographic signals in the form of piecewise stationary random process and developed methods of statistical processing for the problem of compensation of impaired human communicative function.A package of computer programs for statistical processing of electroencephalographic and electromyographic signals has been created on the basis of developed methods that are suitable for use as part of specialized software for automated computer systems for restoring human communicative function.

У дисертації розв'язано актуальну наукову задачу обґрунтування вибору математичної моделі та розроблення методів опрацювання біосигналів, які дають можливість виділення інформативних ознак намагання реалізувати пацієнтами комунікативну функцію в структурі електроенцефалографічних та електроміографічних сигналів для задачі компенсації порушеної комунікативної функції людини.Обґрунтовано вибір математичної моделі таких біосигналів у вигляді кусково стаціонарного випадкового процесу. Розроблено методи опрацювання таких сигналів в межах трансляцій ковзного вікна, для виявлення часових моментів початку та закінчення процесу мовлення а також виявлення ознак наявності основного тону в структурі цих сигналів під час намагання реалізації порушеної комунікативної функції. Обґрунтовано вибір інформативних ознак початку та закінчення процесу мовлення та наявності основного тону.^UIn the dissertation the actual scientific problem of substantiation of the choice of mathematical model and development of methods for processing biosignals, which allow to select informative features of individual voice phonemes in the structure of electroencephalographic and electromyographic signals to compensate for impaired human communication function, is solved.Based on the biophysical foundations of the communicative function, a method of its compensation in cases of its violations is proposed. The method is based on the selection and processing of electroencephalographic signals taken from the surface of the patient's head near the speech centers of the brain, and electromyographic signals taken from the surface of the neck near the vocal folds.A new application of a piecewise stationary random process as a mathematical model of electroencephalographic and electromyographic signals is substantiated, which is adequate to the physical nature of such signals and has the means to highlight informative features that are indicators of the process of communicative function. Based on a well-founded mathematical model, methods for statistical processing of electroencephalographic and electromyographic signals within a sliding window are developed, which are based on calculation and subsequent estimation of power spectral density distributions within sliding window translations and allow to detect time moments signals) and identify signs of the presence of the main tone (based on the results of processing electromyographic signals) during the implementation of human communicative function. It is established that the obtained values of variation of the average estimates of the power spectral density distribution within the sliding window are informative signs of the beginning and end of the speech process, and the average estimates of the electromyographic signal power spectral density distribution in the range of the fundamental frequency are calculated. signs of the presence of the main tone.The structure of the system of selection of electroencephalographic and electromyographic signals is offered. Such signals were selected using the Neurocom electroencephalographic complex as well as the electromyographic signal selection unit, which is a single channel for amplifying the biopotentials of the Neurocom encephalograph with extended bandwidth. Due to the significant noise level in the structure of electromyographic signals, their filtration was performed using the method of spectral subtraction.The results of further processing of the selected biosignals confirmed the correctness of the choice of mathematical model of electroencephalographic and electromyographic signals in the form of piecewise stationary random process and developed methods of statistical processing for the problem of compensation of impaired human communicative function.A package of computer programs for statistical processing of electroencephalographic and electromyographic signals has been created on the basis of developed methods that are suitable for use as part of specialized software for automated computer systems for restoring human communicative function.