基于时频聚集性特征的机械臂控制故障信号远程检测  

Remote Detection of Control Fault Signals in Robotic Arms Based on Time-frequency Clustering Characteristics

作　　者：张剑[1] 欧阳陵江[2] 易守华 ZHANG Jian;OUYANG Lingjiang;YI Shouhua(College of Intelligent Manufacturing,Hunan Network Engineering Vocational College,Changsha 410004,China;College of Mechanical Engineering,Hunan Industry Polytechnic,Changsha 410208,China;Modern Engineering Training Center,Hunan University,Changsha 410082,China)

机构地区：[1]湖南网络工程职业学院智能制造学院,长沙410004 [2]湖南工业职业技术学院机械工程学院,长沙410208 [3]湖南大学现代工程训练中心,长沙410082

出　　处：《机械设计与研究》2024年第6期290-295,302,共7页Machine Design And Research

基　　金：湖南省教育厅科学研究项目(19C1243);湖南省教育科学“十四五”规划项目(XJK21CZJ054);湖南开放大学科研课题(XDK-2021-C-7)。

摘　　要：在实际的工业应用中,为及时发现早期机械臂控制故障,需要对机械臂的关键部位进行实时的在线数据采集、传输和信号处理。但是现场传感器的检测信号传送到控制机房的过程中,当前的故障信号会产生数据干扰、误码、丢失等问题,无法保证云平台远程采集与检测的准确性。为此,提出一种基于时频聚集性特征的机械臂控制故障信号远程检测方法。基于广义S变换正反变换分析机械臂控制故障信号,设计指数函数和调节参数的改进阈值函数,分析机械臂远程控制故障信号的时频分布。得出机械臂产生故障后,验证有效故障信号分量具有时频聚集性。针对这一结论,结合小波分析与维纳滤波方法检测控制故障信号,实现机械臂控制故障信号检测。实验结果表明,所提方法的机械臂各个关节瞬时偏心结果在(0~300)rad范围内保持较为恒定,且各关节在转动过程中,谐波分量维持在(0~20)Hz之间,控制故障信号远程检测精确度维持在98%左右,具有较好的检测能力,建议在机械故障检测中采用。In practical industrial applications,in order to timely detect the control faults of the robotic arms,real-time online data acquisition,transmission and signal processing of the key parts of the arms are needed.However,when the detection signal of the field sensor is transmitted to the control room,the current fault signal will cause data interference,error code,loss and other problems,which cannot guarantee the accuracy of remote collection and detection of the cloud platform.Therefore,remote detection of robotic arm control fault signals is proposed based on time-frequency clustering features.Based on the the generalized S transform of the control fault signal,the improved threshold function of the index function and the regulation parameters are designed,and the time-frequency distribution of the remote control fault signal is analyzed.After the mechanical arm is faulty,it is verified that the effective fault signal component has time-frequency aggregation.According to this conclusion,the wavelet analysis and Wiener filtering method are used to detect control fault signal and realize mechanical arm control fault signal detection.The experimental results show that the instantaneous eccentricity results of each joint of the proposed method's robotic arm remain relatively constant within the range of 0~300 rad,and the harmonic components of each joint are maintained between 0~20 Hz during rotation.The remote detection accuracy of control fault signals is maintained at around 98%,indicating good detection ability.It is recommended to use this method in mechanical fault detection.

关 键 词：小波包变换 机械臂 控制故障信号 时频聚集 广义S变换 维纳滤波 

分 类 号：TH137[机械工程—机械制造及自动化]