DEVELOPMENT OF A MATHEMATICAL MODEL FOR NONLINEAR ELECTRIC DRIVES OF EXOSKELETON AND SUBORDINATE SELF-TUNING CONTROL USING A NEURAL NETWORK

Proposes a subordinate controller with self-tuning adaptation of the position control loop coefficients by an analog neural network for a nonlinear electric drive of an exoskeleton. An excellent blend of slave controller and neural network with powerful continuous learning, adaptation and nonlinearity problem solving capabilities, it provides a new nonlinear neural network controller based on proportional integral differential (PID) control methods. To correct the coefficients of traditional PID controllers, it is proposed to use a neural network (NN), which allows self-tuning the coefficients directly at its outputs based on the errors of the control system. The development of a dynamic model of an exoskeleton includes five links of two legs and a body, taking into account non-linear electric drives as a control object. The nonlinear approach to modeling is due to the use of a control system for the electric drives of the exoskeleton under various conditions, which is multi-mass rotating in two directions with nonlinearities of the form of viscous and Coulomb friction, elasticity and external disturbance. A method of subordinate regulation of exoskeleton nonlinear electric drives with a neural network for tracking control of the trajectory of exoskeleton joints is presented. Also, the results of the experiment on the MatLab/Simulink program were obtained to illustrate the effectiveness of the proposed control strategy.

Authors: D. D. Truong, M. P. Belov

Direction: Electrical Engineering

Keywords: Exoskeleton, electric drive, slave regulator, self-tuning, neural network, mathematical model


View full article