Double Iterative Compensation Learning Control for Active Training of Upper Limb Rehabilitation Robot

Abstract

In this paper, the problem of non parametric uncertainty in the active training stage of stroke patients is discussed. On the basis of the nonlinear iterative learning theory, a double iterative compensation learning control is proposed. This method adopts the strategy of double loop iteration which can adjust controller parameters in real time to satisfy the patients’ condition. First, a class of saturated nonlinear functions is introduced to satisfy the requirement of position constraints. Then, reference trajectory self-modified strategy is designed for the initial positioning error. Meanwhile, the iterative compensation controller is designed according to the active torque of the patient, which can provide appropriate power compensation to the affected limb and update the parameters of the iterative learning controller continuously. At last, the convergence condition and its proof are given. The simulation results show the effectiveness and practicability of the proposed double iterative compensation optimal control strategy.