Optimization of Knee Exoskeleton Rehabilitation Trajectory Based on Iterative Learning
Abstract
In the process of knee rehabilitation, the traditional control method has a fixed trajectory and cannot balance the early high-precision tracking and the later human-machine compliance. This paper proposes an iterative learning control (ILC) scheme based on the dynamic adjustment mechanism of phased learning rate. To begin with, A single degree-of-freedom kinetic model of the knee exoskeleton was constructed using the subject's anatomical data. Subsequently, this paper designs a stepwise renewal law that uses a high learning rate to accelerate the establishment of motor memory during the passive rehabilitation period, and a stepwise downward adjustment of the learning rate during the active rehabilitation period to enhance the compliance of the system to the body's active force. The simulation results show that the fluctuation of root mean square error (RMSE) under active interference is much smaller than that of the traditional fixed gain algorithm. Microscopic analysis further confirmed that the strategy could effectively inhibit the torque distortion caused by human-machine confrontation, and while ensuring the stability of the rehabilitation trajectory, it gave the system good physical flexibility, which provided theoretical support for full-cycle personalized rehabilitation.