坐卧式下肢康复机器人的被动训练控制  

Passive Training Control of Horizontal Lower Limbs Rehabilitative Robot

作　　者：吕显耀 杨炽夫[1] 姜峰[2] 韩俊伟[1] LV Xian-yao;YANG Chi-fu;JIANG Feng;HAN Jun-wei(School of Mechatronics Engineering Harbin Institute of Technology, Heilongjiang Harbin 150001, China;School of Computer Science and Technology Harbin Institute of Technology, Heilongjiang Harbin 150001, China)

机构地区：[1]哈尔滨工业大学机电工程学院,黑龙江哈尔滨150001 [2]哈尔滨工业大学计算机科学与技术学院,黑龙江哈尔滨150001

出　　处：《机械设计与制造》2019年第4期244-247,共4页Machinery Design & Manufacture

基　　金：国家自然科学基金资助项目(51305095)

摘　　要：考虑到下肢康复机器人很难获得精确、完整的数学模型,而且在建立模型时需要进行合理的近似处理,因此忽略了外部干扰、参数误差、未建模的动态和摩擦等不确定因素,这些原因导致控制性能不佳。基于此提出了一种基于计算力矩法的神经网络鲁棒控制器,通过计算力矩法对标称模型进行控制,RBF神经网络控制器对系统中未知的不确定项进行补偿,而自适应鲁棒控制器则用来补偿神经网络的逼近误差及外部的干扰,从而提高了系统的动态性能和控制精度,并对算法的稳定性进行了证明。通过实验验证,证明了控制算法的有效性,在被动训练时具有较好的轨迹跟踪性能。It is difficult to obtain an accurate and complete mathematical model for lower limbs rehabilitative robot, and we will do some reasonable approximate treatments when building the model, so the external disturbance, parameter error, unmodeled dynamics and friction are ignored. These reasons will cause poor control performance. The neural network robust controller based on the computed torque method is presented in this paper. The ideal dynamic model is controlled by the computed torque method, and RBF neural network controller compensates for the unknown uncertainties, then the adaptive robust controller compensates the approximation error of neural network and external interference. So the algorithm will improve the system dynamic performance and control accuracy. And the stability of the algorithm is proved in this paper. The experimental results show that the control algorithm is effective and has good trajectory tracking performance when do the passive training.

关 键 词：外骨骼 康复机器人 轨迹跟踪 被动训练 

分 类 号：TH16[机械工程—机械制造及自动化] TP242[自动化与计算机技术—检测技术与自动化装置]