摘要
针对理疗师劳动强度较大的问题,提出一种下肢康复机器人的解决方案.运用循环神经网络对机器人进行建模,并利用该模型预测机器人的关节状态,采用牛顿-拉夫逊优化算法,根据预测误差进行反馈校正,实现对机器人位置的预测控制.仿真结果表明该控制方法能够满足康复机器人的控制要求.
Aiming at the problem of labor intensity of physiotherapists,a solution for rehabilitation robot of lower limbs is proposed.The robot is modeled by a recurrent neural network and the joint state of the robot is predicted by this model.The optimization is implemented by Newton-Raphson algorithm,and the feedback correction is based on prediction error.Using this,it is possible to predictively control the position of the robot.The simulation results show that the proposed control strategy can meet the requirements of the rehabilitation robot control.
出处
《扬州大学学报(自然科学版)》
CAS
北大核心
2017年第3期44-48,57,共6页
Journal of Yangzhou University:Natural Science Edition
基金
国家自然科学基金资助项目(61303183)
江苏省教育厅科学技术计划项目(15KJB120010)
徐州工程学院培育项目(XKY2015202)
关键词
神经网络
预测控制
下肢康复
机器人
neural network
predictive control
lower-limb rehabilitation
robot