一种模型与数据混合驱动的机器人柔性关节振动分离方法  

A vibration separation method for flexible joints of robot driven by combination of model and data

作　　者：李健龙 柳小勤[1,2] 伍星[1,2,3] 王东晓 徐凯 LI Jianlong;LIU Xiaoqin;WU Xing;WANG Dongxiao;XU Kai(Faculty of Mechanical and Electrical Engineering,Kunming University of Science and Technology,Kunming 650500,China;Yunnan Provincial Engineering Research Center for Advanced Equipment Intelligent Maintenance,Kunming University of Science and Technology,Kunming 650500,China;Faculty of Mechanical Engineering,Yunnan Vocational College of Mechanical and Electrical Technology,Kunming 650203,China)

机构地区：[1]昆明理工大学机电工程学院,昆明650500 [2]昆明理工大学云南省先进装备智能维护工程研究中心,昆明650500 [3]云南机电职业技术学院机械工程学院,昆明650203

出　　处：《振动与冲击》2024年第23期12-19,共8页Journal of Vibration and Shock

基　　金：国家重点研发计划项目课题(2018YFB1306103);云南省重大科技专项计划(202202AC080008)。

摘　　要：工业机器人关节具有柔性,会产生较大的工作振动。针对机器人关节发生故障时,如何从测量得到的混合振动信号中分离出故障分量的问题,提出一种模型与数据混合驱动的关节振动分离方法。首先,建立多物理量信号和系统动力学相结合的执行器动力学响应模型,并以该响应信号作为振动分离时的参考。其次,考虑噪声影响构造了幅值谱百分位序列,利用变点分析确定最优噪声阈值,并设计带通滤波器分离噪声。针对测量和滤波带来的参考振动和混合振动间相位误差问题,提出可调节因子动态时间规整相位校正方法。最后,由去噪和相位校正后的混合振动减去参考振动实现故障分量分离。在机器人关节试验台上的试验结果表明,所提方法能从关节振动中有效地分离出故障分量。Joints of industrial robots are flexible and can generate larger working vibrations.Here,aiming at the problem of how to separate fault components from the measured mixed vibration signals when robot joints having faults,a model and data-driven joint vibration separation method was proposed.Firstly,an actuator dynamic response model combining multi-physical quantity signals with system dynamics was established,and this response signal was taken as a reference for vibration separation.Secondly,considering effects of noise,an amplitude spectral percentile sequence was constructed,and the optimal noise threshold was determined using variable point analysis.A bandpass filter was designed to separate noise.Aiming at the phase error problem between reference vibration and mixed vibration caused by measurement and filtering,an adjustable factor dynamic time warping phase correction method was proposed.Finally,fault components separation was realized by subtracting reference vibration from the denoised and phase-corrected mixed vibration.The experimental results on a robot joint test platform showed that the proposed method can effectively separate fault components from joint vibrations.

关 键 词：机器人关节故障 故障分离 可调节因子动态时间规整 噪声分离 数据生成 

分 类 号：TH165.3[机械工程—机械制造及自动化] TP242.2[自动化与计算机技术—检测技术与自动化装置]