基于非线性摩擦力模型和WLS-CS算法的机器人动力学参数辨识  

Dynamic Parameter Identification for Robot Based on Nonlinear Fiction Model and WLS-CS Algorithm

作　　者：梁艳阳 黄子健 孙伟霖 石峰 钟铭溪 Liang Yanyang;Huang Zijian;Sun Weilin;Shi Feng;Zhong Mingxi(Department of Intelligent Manufacturing,Wuyi University,Jiangmen,Guangdong 529020,China)

机构地区：[1]五邑大学智能制造学部,广东江门529020

出　　处：《机电工程技术》2024年第9期169-172,共4页Mechanical & Electrical Engineering Technology

基　　金：五邑大学科研启动基金项目(409170190241)。

摘　　要：为了实现对工业机器人的高精度控制,需要一个高精度的动力学模型,因此要对动力学参数进行精准辨识。提出一种基于非线性摩擦力模型和加权最小二乘法与布谷鸟算法(WLS-CS)结合的动力学参数辨识方法。首先通过Newton-Euler法进行机器人的动力学建模,结合非线性摩擦力模型,完成机器人全动力学模型的建模。设计5阶有限项傅里叶级数作为激励轨迹,并优化激励轨迹参数。使用加权最小二乘法对动力学参数进行初步辨识。辨识结果作为布谷鸟算法的初始解范围,结合Levy飞行与巢穴的优胜劣汰机制,寻找到动力学参数辨识的全局最优解。更换轨迹对机器人全动力学模型进行验证,计算机器人的仿真力矩,并与实验力矩反馈值进行对比。结果表明上述方法辨识的动力学参数辨识精度的残差方根(RMS)值下降8.6%,可以建立精确度较高的动力学模型。In order to improve the identification accuracy of dynamic parameters of industrial robots,an identification method based on nonlinear friction model,weighted least square method and Cuckoo algorithm(WLS-CS)is proposed.Firstly,the robot dynamics model is built by Newton-Euler method,combined with the nonlinear friction model,and the robot dynamics model is completed.The fifth order finite Fourier series is designed as the excitation locus.The kinetic parameters are initially identified by weighted least square method.The identification results are taken as cuckoo nests,combined with Levy flight and the survival of the fittest mechanism of nests,and the global optimal solution of dynamic parameter identification is found.The full dynamic model is verified by replacing the trajectory,and the feedback values of the simulated torque and the experimental torque are compared.The results show that the residual root mean square(RMS)value of the dynamic parameter identification accuracy identified by the above method decreases by 8.6%,and a high-precision dynamic model can be established.

关 键 词：动力学参数辨识 工业机器人 非线性摩擦力模型 最小二乘法 布谷鸟算法 

分 类 号：TP242.2[自动化与计算机技术—检测技术与自动化装置]