摘要
设计了一种气动人工肌肉驱动的串联弹性关节,基于气动人工肌肉Chou模型,建立了串联弹性关节的动力学模型,推导出关节刚度,获得了关节刚度与肌肉内部气压、弹性体刚度的关系。设计反向传播(back propagation,BP)神经网络整定PID参数的BP-PID控制算法,开展了气动串联弹性关节的位置与刚度控制研究。仿真结果表明BP-PID控制优于PID控制,关节位置跟踪精度由0.58°变为0.10°,关节刚度跟踪精度从0.026 N·m/rad变为0.005 N·m/rad。实验结果表明关节位置跟踪平均误差由0.347°减小到0.117°,关节刚度跟踪平均误差从0.024 N·m/rad减小到0.019 N·m/rad。
A kind of series elastic joints driven by pneumatic artificial muscle(PAM)was proposed.Based on the Chou model of PAM,a dynamic model of series elastic joints was established,and joint stiffness was derived.The relationship between the joint stiffness and internal pressure of PAM and stiffness of elastomer was obtained.The control algorithm of BP neural network tuning PID parameters(BP-PID)was designed,and the research on position and stiffness control of pneumatic series elastic joints was performed.The simulation results show that BP-PID control is better than PID control,tracking errors of joint positions are changed from 0.58°to 0.10°,and tracking errors of joint stiffness are changed from 0.026 N·m/rad to 0.005 N·m/rad.The experimental results show that the average tracking error of planning position signal is reduced from 0.347°to 0.117°,and the average tracking error of joint stiffness is reduced from 0.024 N·m/rad to 0.019 N·m/rad.
作者
沈双
雷静桃
张悦文
SHEN Shuang;LEI Jingtao;ZHANG Yuewen(School of Mechatronic Engineering and Automation,Shanghai University,Shanghai,200444)
出处
《中国机械工程》
EI
CAS
CSCD
北大核心
2021年第12期1486-1493,共8页
China Mechanical Engineering
基金
国家自然科学基金(51775323,51375289)。
关键词
仿生跳跃机器人
串联弹性驱动器
反向传播神经网络
位置控制
刚度控制
bionic hopping robot
series elastic actuator
back propagation(BP)neural network
position control
stiffness control
