摘要
为解决爬壁机器人轨迹跟踪过程中速度突变与输入抖振的问题,提出一种基于神经动力学模型的反演运动学控制器与组合趋近律神经滑模动力学控制的混合鲁棒控制算法。利用神经动力学模型获取有界、平滑的虚拟位姿误差信号,解决了传统反演控制法引起的速度跳变问题;引入自适应径向基神经网络(RBFNN)调节基于组合趋近律的滑模增益,消除了抖振现象。设计过程采用Lyapunov函数,保证了控制系统的稳定与收敛。通过仿真数据与实验结果证明了所提算法的有效性。
To solve the problem of velocity jump and input chattering in the track trajectory of the wall-climbing robot,a hybrid robust control algorithm was proposed based on a backstepping kinematics controller of neuro dynamics model and a neural sliding mode dynamics controller of combined asymptotic law.The bounded and smooth virtual posture errors were obtained by using the neuro dynamics model to suppress the sharp velocity jump caused by the traditional backstepping algorithm.The gain of sliding mode controller based on a combined asymptotic law was adjusted by using the adaptive radial basis function neural network to eliminate input chattering.The simulation data and experimental results showed the effectiveness of the proposed algorithm.
作者
张小俊
吴亚淇
谢必成
赵金亮
ZHANG Xiaojun;WU Yaqi;XIE Bicheng;ZHAO Jinliang(School of Mechanical Engineering,Hebei University of Technology,Tianjin 300130,China)
出处
《计算机集成制造系统》
EI
CSCD
北大核心
2023年第11期3560-3571,共12页
Computer Integrated Manufacturing Systems
基金
国家重点研发计划资助项目(2018YFB1309401)。
关键词
爬壁机器人
轨迹跟踪
混合控制
神经动力学模型
反演控制
神经滑模控制
wall climbing robot
trajectory tracking
hybrid control
neuro dynamics model
backstepping control
neural sliding mode control
