SELF-TUNING PID CONTROLLER BY NEURAL NETWORK FOR NONLINEAR ELECTRIC DRIVE SYSTEM OF THE EXOSKELETON

This article proposes a proportional integral derivative (PID) controller with coefficients corrective adaptation by neural network for a nonlinear electric drive of the exoskeleton. Unlike PID controllers with fixed values of the coefficients, which are not capable of adapting the model parameters under operating conditions, self-tuning of the PID controller with neural network is a new direction for the nonlinear electric drive of the exoskeleton. To correct the coefficients of traditional PID controllers, it is proposed to use neural network with a radial basis function (RBF), which allows the coefficients to be self-adjusted directly from its outputs based on the control system errors. The control system of the exoskeleton electric drive is a complex system that requires high accuracy and provides the disabled with stable movement. The features of automated control systems for exoskeleton electric drives are the presence of gaps in the kinematic transmissions, significant moments of friction in the moving parts of the structure. A modern approach to solving such problems is the use of neural network with self-tuning of the PID controller coefficients. The simulation results show that the controller has strong adaptability compared to the conventional PID controller.

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

Direction: Electrical Engineering

Keywords: Neural network, exoskeleton, electric drive, PID controller, self-tuning, control system

---

View full article