Adaptive Neural Network Control of Lower Limb Exoskeleton Robots Using Disturbance Observer

摘要

This paper raises an adaptive neural network control method for motion tracking of a lower limb rehabilitation exoskeleton robot. This method is based on RBFNN and disturbance observer. The motion tracking is hard to control since it is highly nonlinear, and suffers from large uncertainties. Fitting by the neural network, the dynamic model is linearized. The adaptive linear model is proved to be stable and an adaptive output is found. A bounded equivalent form of the model is obtained based on the adaptive output. RBFNN is adopted to guarantee both control accuracy and strong robustness, and a disturbance observer is used to eliminate the effect of the unknown disturbance online. Experiment results are presented and the results show the validity of the control strategies.