全状态约束下长行程混联机器人投影迭代鲁棒控制算法  

Projection iterative robust control algorithm for long-stroke hybrid robot under full-state constraints

作　　者：刘群坡[1,2] 张卓然 张建军 卜旭辉 孙蕊[1] LIU Qunpo;ZHANG Zhuoran;ZHANG Jianjun;BU Xuhui;SUN Rui(School of Electrical Engineering and Automation,Henan Polytechnic University,Jiaozuo 454003,China;International Joint Laboratory of Direct Drive and Control of Intelligent Equipment,Jiaozuo 454003,China)

机构地区：[1]河南理工大学电气工程与自动化学院,河南焦作454003 [2]河南省智能装备直驱技术与控制国际联合实验室,河南焦作454003

出　　处：《兵器装备工程学报》2024年第S01期322-332,共11页Journal of Ordnance Equipment Engineering

基　　金：国家自然科学基金项目(U1804147);河南省高校科技创新团队项目(20IRTSTHN019);河南省科技攻关项目(212102210508)。

摘　　要：针对全状态约束下的长行程混联机器人系统鲁棒性较差提出了基于自适应学习神经网络和等效误差函数的投影迭代鲁棒控制算法。基于自适应学习神经网络逼近未知的非线性项,提出投影迭代鲁棒控制算法,更新网络权值并估计逼近误差和随机外部扰动的未知上界;构造用于抵消初始时刻随机变化扩展误差的时变边界层,设计基于时变边界层和扩展误差的等效误差函数作为迭代控制器的主要控制变量以克服随机初始误差满足相同初始条件;在控制器设计中引入正切型障碍Lyapunov函数,确保系统状态在预定范围内。仿真实验结果证明了该方法的有效性,可在全状态约束下实现高精度强鲁棒性的轨迹跟踪。A projection iterative robust control algorithm based on adaptive learning neural network and equivalent error function was proposed for a long-stroke hybrid robot under full state constraints,when the system was poorly robust due to the simultaneous presence of uncertainty and random initial errors.Based on adaptive learning neural networks to approximate unknown nonlinear terms,a projection iterative robust control algorithm was proposed to update network weights and estimate the unknown upper bound of approximation errors and random external disturbances.To overcome the random initial errors and satisfy the same initial condition,a time-varying boundary layer was constructed to offset the random variation expansion error at the initial time,and an equivalent error function based on the time-varying boundary layer and expansion error was designed as the main control variable of the iterative controller.The tangent-type barrier Lyapunov function was introduced into the controller design to ensure that the system state was within a predetermined range.Finally,the simulation experimental results verify the effectiveness of the method,which can achieve high-precision and strong robust trajectory tracking under full-state constraints.

关 键 词：自适应迭代学习控制 长行程混联机器人 神经网络 随机初始误差 状态约束 

分 类 号：TP242.2[自动化与计算机技术—检测技术与自动化装置]