Fuzzy neural network and observer-based fault-tolerant adaptive nonlinear control of uncertain 5-DOF upper-limb exoskeleton robot for passive rehabilitation

Abstract

This paper investigates the control of a 5-DOF upper-limb exoskeleton robot used for passive rehabilitation therapy. The robot is subject to uncertain dynamics, disturbance torques, unavailable full-state measurement, and different types of actuation faults. An adaptive nonlinear control scheme, which uses a new reaching law-based sliding mode control strategy, is proposed. This scheme incorporates a high-gain state observer with dynamic high-gain matrix and a fuzzy neural network (FNN) for state vector and nonlinear dynamics estimation, respectively. Using dynamic parameters, the scheme provides an efficient mean for simultaneously tackling the effects of FNN approximation errors, disturbance torques and actuation faults without any prior bounds knowledge and fault detection and diagnosis components. Using simulation results, it is shown that with the presented scheme, faster response, fewer oscillations during transient phase, good tracking accuracy, and chattering-free control torques with lower amplitudes are obtained.