实时预测下的机器人关节连续转动角度误差前置性补偿方法  

Pre Compensation Method for Continuous Rotation Angle Error of Robot Joints Under Real-time Prediction

作　　者：曹丽芳 岳鹏 张建忠[1] CAO Lifang;YUE Peng;ZHANG Jianzhong(Department of Mechanical Engineering,Yellow River Conservancy Technical Institure,Kaifeng Henan 475004,China;College of Chemistry and Molecular Sciences,Henan university,Zhengzhou 450046,China)

机构地区：[1]黄河水利职业技术学院机械工程学院,河南开封475004 [2]河南大学化学与分子科学学院,郑州450046

出　　处：《机械设计与研究》2024年第3期36-40,45,共6页Machine Design And Research

基　　金：河南省教育厅科技攻关项目(202102210265)。

摘　　要：由于机器人关节连杆之间存在非线性、耦合效应以及定位误差,当前的机器人关节误差补偿方法都是后置性的,以产生误差-误差校准的思路完成补充。一旦产生误差过大,就会造成实际损失。提出实时预测下的机器人关节连续转动角度误差补偿方法。利用修正的完整参数连续(Modified Complete Parameter Continuity, MCPC)方法,建立惯性坐标系,使机器人的关节运动满足时间序列上的约束条件。引入卡尔曼滤波算法,并对关节空间内的机器人关节转动状态展开预测及更新;根据最终转动角度更新结果,与期望转动角度的对比,实现实时转角误差判断,不再依赖后置性误差结果。根据模糊控制器论域与转角误误差之间的自适应变化关系,建立模糊规则库作为补偿历史数据,实现关节连续转动角度误差的主动前置性补偿。实验证明,所提方法在机器人关节转动中,能够精准识别当前转动角度,误差补偿效果明显,对机器人作业精度的提升具有一定意义。Due to the nonlinearity,coupling effects and positioning errors between robot joint linkages,current robot joint error compensation methods are all post positional,supplemented by the idea of error calibration after errorgeneration.Once the error is too large,it will cause actual losses.A real-time prediction based compensation method is proposed for continuous rotation angle error of robot joints.Using the modified complete parameter continuity(MCPC)method,an inertial coordinate system is established to ensure that the joint motion of the robot meets the constraints in the time series.Introducing Kalman filtering algorithm,integrating multi-source kinematic data,predicting and updating the rotational status of robot joints in joint space,and based on the updated results of the final rotation angle and the comparison with the expected rotation angle,the real-time angle error judgment is achieved,without relying on post position error results.Based on the adaptive variation relation between the domain of fuzzy controller and angle error,a fuzzy rule library is established as compensation historical data to achieve active proactive compensation for joint continuous rotation angle error.Experiments have shown that the proposed method can accurately identify the current rotation angle in robot joint rotation,and the error compensation effect is obvious,which has certain significance for improving the accuracy of robot operations.

关 键 词：MCPC方法 运动学模型 关节空间 模糊规则库 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]