CASPIAN JOURNAL
MANAGEMENT AND HIGH TECHNOLOGIES
TWO-LEVEL NEURAL NETWORK MODEL OF ELECTROMIOSIGNAL DECODER IN EXOSCELETE VERTICALIZATION CONTROL SYSTEM
 |
Read | Trifonov Andrey A., Filist Sergey A., Kuzmin Alexander A., Zhilin Valery V., Petrunina Elena V. TWO-LEVEL NEURAL NETWORK MODEL OF ELECTROMIOSIGNAL DECODER IN EXOSCELETE VERTICALIZATION CONTROL SYSTEM // Caspian journal : management and high technologies. — 2020. — №4. — pp. 99-111. |
Trifonov Andrey A. - Southwest State University, voldraf@mail.ru
Filist Sergey A. - Southwest State University, sfilist@gmail.com
Kuzmin Alexander A. - Southwest State University, ku3bmin@gmail.com
Zhilin Valery V. - Kursk Institute of Cooperation, a branch of the Belgorod University of Consumer Cooperation, Economics and Law, vvzhilin61@gmail.com
Petrunina Elena V. - Moscow State University of the Humanities and Economics, petrunina@mggeu.ru
An exoskeleton control system in a biotechnological system designed to restore the motor activity of a patient’s muscles is presented. The essence of the exoskeleton control method used is that when decoding the electromiosignal, not only its amplitude indicators, but also frequency characteristics are used, since it is known that an increase in motor activity leads not only to an increase in the amplitude of the electromiosignal, but also to an increase in the number of motor units involved. Also, as in the known methods, in order to adapt the patient’s functional state and the verticalization process, the decoder uses many duplicate channels of electromiosignals associated with the muscle or muscle groups that control the movement of the same limb joint, resulting in the output of each classifier of the channel, we obtain the number corresponding to the confidence in the control command of the exoskeleton servomotor, and for aggregation of decisions on the channels of classifiers, all outputs of the channel classifiers go to a fuzzy neural network, the defusifier of which generates a control signal to the servo motor controller, as a result of which analysis the controller determines the speed and direction of rotation. Based on the basic models of the classifier, the exoskeleton was controlled in the “stand up - sit down” mode. The location of the electrodes on the muscle group for the implementation of the verticalization mode with combined control is determined. It is shown that a simplified kinematic model of the verticalization mode allows, together with a two-level neural network model of an electromyosignal decoder, to adapt the rehabilitation process to the functional state of the patient. The obtained models of classifiers of surface signals of electromyograms can be used to build intelligent rehabilitation systems for patients with neurological diseases and will allow the development of adaptive stimulating programs, testing the results of which will allow developing new methods and tools for rehabilitation of patients with neurological diseases.
Key words: electromyosignal, exoskeleton, decoder of electromyosignal, neural network, solution aggregator, algorithm, biotechnical system, adaptation of the rehabilitation system to the patient