基于多变量预测补偿的机械臂精度提升方法  

Promoting method for manipulator accuracy based on multi-variable prediction and compensation

作　　者：王琨[1,2,3] 骆敏舟[3] 曹毅[1,2] 李可[1,2] 张秋菊[1,2] 

机构地区：[1]江南大学江苏省食品先进制造装备技术重点实验室,无锡214122 [2]江南大学机械工程学院,无锡214122 [3]中国科学技术大学信息科学技术学院,合肥230027

出　　处：《电子测量与仪器学报》2014年第11期1213-1221,共9页Journal of Electronic Measurement and Instrumentation

基　　金：国家自然科学基金(50905176);中央高校基本科研业务费专项(JUSRP11456);江苏省食品先进制造装备技术重点实验室开放课题(FM-2014-05)资助项目

摘　　要：为了减小由运动学参数不准确引起的机械臂位姿误差,传统方法是标定机械臂的连杆参数,但标定后的真实连杆参数不便于逆运动学解析解计算。提出一种基于遗传算法的对机械臂关节旋转变量直接进行修正的方法来提来机械臂定位精度。首先建立机械臂运动学模型和误差模型,得到关节修正量的求解公式。通过NDI三维动态位移测量系统检测机械臂末端执行器的实际位姿,作为样本,并构建遗传算法的适应度函数,采用遗传算法计算出各关节旋转变量补偿量。最后,将该修正值应用于设计的六自由度串联机械臂的控制,多组实验数据证明该方法的有效性及其提高机械臂的绝对定位精度的性能。The traditional parameter identification techniques are effective for decreasing the position and orientation errors of the end-effector which are induced by the inaccurate kinematics parameters of the manipulator.However, when the identified physical parameters are applied, the mathematical calculation of the inverse kinematics the ma-nipulator becomes complicated.In this paper, a compensation method of adjusting the joint rotation variables of the manipulator based on the genetic algorithms is presented for improving the model-based control accuracy.Firstly, the kinematics model and the error model of the manipulator are built, and the solution formula of variables com-pensation is obtained.Then the actual positions and the postures of the manipulator which are measured by the NDI 3D dynamic displacement measurement system are used as experimental samples.Moreover, the fitness function of the genetic algorithms is built, and the best joint variables compensations are computed.Finally, the variables compensations are applied for the control of the designed 6-DOF serial manipulator, and numerical examples are used to demonstrate the effectiveness of the method, as well as the performance of the improvement of the positio-ning accuracy.

关 键 词：遗传算法 串联机械臂 运动学模型 参数辨识 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]