This article presents a novel adaptive iterative learning control (AILC), and designs a human-in-loop control pattern (HIL-CP), which simulates the proposed approach using different lower limb rehabilitation robot models. The stability of the AILC controller is proposed and verified via a Lyapunov-like function, where novel controller shows strong robustness in disturbances environment. Based on AILC, the core of the HIL-CP interactive control mode is to estimate the human surface electromyography by neural network model and get the real-time desired trajectory to iterate out the optimal actual tracking trajectory, which reduce the tracking error quickly and ensure the rehabilitation training effect of patients. Furthermore, the MATLAB software is employed to conduct simulation experiments the proposed approach. The simulation results show that the HIL-CP is highly efficient and rapidly convergent in a satisfied degree. The angle error is \({\mathrm{{0.25}}^\text {o}}\pm {\mathrm{{0.2}}^\text {o}} \) for patients and \({\mathrm{{0.03}}^\text {o}}\pm {\mathrm{{0.02}}^\text {o}} \) for healthy people. Compared with the existing sliding mode controller, it is proven that the AILC controller is much more effective and noise-tolerant ability in the presence of bounded nonlinear disturbance.

本文提出了一种新颖的自适应迭代学习控制 (AILC)，并设计了一种人在环控制模式 (HIL-CP)，该模式使用不同的下肢康复机器人模型来模拟所提出的方法。通过类李雅普诺夫函数提出并验证了 AILC 控制器的稳定性，其中新型控制器在扰动环境中表现出很强的鲁棒性。基于AILC的HIL-CP交互控制模式的核心是通过神经网络模型对人体表面肌电进行估计，得到实时的期望轨迹，迭代出最优的实际跟踪轨迹，快速降低跟踪误差，确保患者康复训练效果。此外，采用MATLAB软件对所提出的方法进行仿真实验。仿真结果表明，HIL-CP 高效且快速收敛。角度误差为\({\mathrm{{0.25}}^\text {o}}\pm {\mathrm{{0.2}}^\text {o}} \)用于患者和\({\mathrm{{0.03}}^ \text {o}}\pm {\mathrm{{0.02}}^\text {o}} \)适用于健康人。与现有的滑模控制器相比，证明了AILC控制器在存在有界非线性扰动的情况下更加有效和抗噪能力更强。