工业机器人驱动系统非线性频谱故障诊断方法
|
摘要
针对广义频率响应函数(GFRF)在故障诊断中存在计算量大、无法满足系统对诊断实时性要求的问题,提出基于非线性输出频率响应函数(NOFRF)的工业机器人驱动系统故障诊断方法。该方法构建系统一维频谱函数的辨识模型,将系统的输出频谱与估计频谱进行比较求出残差,根据残差大小改变辨识步长迭代出前4阶频谱;对获取到的4阶频谱进行逐阶采样,每阶频谱采集10个数值,共40个频谱构成40维特征矢量,将其作为系统的故障特征输入核主成分分析方法(KPCA)进行压缩,通过计算主元累计贡献率将高维数据压缩至3维,降低变量之间的非线性度;构造SVM分类器,将KPCA方法生成的低维数据中60%的数据作为训练集对分类器进行训练,将40%的数据作为测试集进行故障识别。实验结果表明,在相同的数据提取任务下,与基于GFRF的方法相比,所提方法节约时间854%,可以准确、快速地提取系统故障特征,进一步验证了该方法在工业机器人驱动系统故障诊断应用上的可靠性。
A fault diagnosis method for industrial robot drive systems based on nonlinear output frequency response function(NOFRF) is proposed to solve the problem that the generalized frequency response function(GFRF) cannot satisfy the real-time requirement of systems due to its large amount of calculation in fault diagnosis. The method constructs an identification model of one-dimensional spectrum function for a system, and calculates the residual between the output spectrum and the estimated spectrum of the system in order to change iteration step length of identification and to get spectrum values of the first four orders. These four orders' NOFRF spectrum values are then sampled order by order with 10 values for each order, and a total of 40 spectrum values form a 40-dimensional feature vector. The resulting vectors are sent to the kernel principal component analysis(KPCA) for compression by calculating the cumulative contribution rate of the principal component. The high-dimensional data are compressed into 3-dimensional data to reduce the nonlinearity between variables. An SVM classifier is constructed, and 60% of the low-dimensional data generated by KPCA are used as a training set to train the classifier, while the rest 40% of the data are used as a test set to identify faults. Experimental results and a comparison with GFRF show that the proposed method saves 85.4% of the processing time under the same data extraction task, and accurately and quickly extracts fault features of the system, which proves the reliability of the method in the application to fault diagnosis of industrial robot drive systems.
A fault diagnosis method for industrial robot drive systems based on nonlinear output frequency response function(NOFRF) is proposed to solve the problem that the generalized frequency response function(GFRF) cannot satisfy the real-time requirement of systems due to its large amount of calculation in fault diagnosis. The method constructs an identification model of one-dimensional spectrum function for a system, and calculates the residual between the output spectrum and the estimated spectrum of the system in order to change iteration step length of identification and to get spectrum values of the first four orders. These four orders' NOFRF spectrum values are then sampled order by order with 10 values for each order, and a total of 40 spectrum values form a 40-dimensional feature vector. The resulting vectors are sent to the kernel principal component analysis(KPCA) for compression by calculating the cumulative contribution rate of the principal component. The high-dimensional data are compressed into 3-dimensional data to reduce the nonlinearity between variables. An SVM classifier is constructed, and 60% of the low-dimensional data generated by KPCA are used as a training set to train the classifier, while the rest 40% of the data are used as a test set to identify faults. Experimental results and a comparison with GFRF show that the proposed method saves 85.4% of the processing time under the same data extraction task, and accurately and quickly extracts fault features of the system, which proves the reliability of the method in the application to fault diagnosis of industrial robot drive systems.
引文
[1]STAVROU D,ELIADES D G,PANAYIOTOU C G,et al.Fault detection for service mobile robots using model-based method[J].Autonomous Robots,2016,40(2):383-394.
[2]MA Hongjun,YANG Guanghong.Simultaneous fault diagnosis for robot manipulators with actuator and sensor faults[J].Information Sciences,2016,366(10):12-30.
[3]陈钢,郭雯,贾庆轩,等.基于运动学模型重构的单关节故障机械臂容错路径规划[J].控制与决策,2018,33(8):1436-1442.CHEN Gang,GUO Wen,JIA Qingxuan,et al.Faulttolerant path planning of single-joint fault manipulator based on kinematics model reconstruction[J].Control and Decision,2018,33(8):1436-1442.
[4]刘冠初,熊静琪,乔林,等.四足机器人自由步态规划建模与算法实现[J].西安交通大学学报,2015,49(6):84-89.LIU Guanchu,XIONG Jingqi,QIAO Lin,et al.Modelling and algorithm realization of free gait regulation for quadruped robots[J].Journal of Xi’an Jiaotong University,2015,49(6):84-89.
[5]盖海松,徐方,贾凯,等.一种机器人关节扭矩传感器的设计与研究[J].现代制造工程,2017(5):33-37.GAI Haisong,XU Fang,JIA Kai,et al.Design and research of a robot joint torque sensor[J].Modern Manufacturing Engineering,2017(5):33-37.
[6]严浩,白瑞林,吉峰.一种改进的SCARA机器人动力学参数辨识方法[J].中国机械工程,2017,28(22):2707-2713.YAN Hao,BAI Ruilin,JI Feng.An improved dynamic parameter identification method for SCARA robots[J].China Mechanical Engineering,2017,28(22):2707-2713.
[7]JABER A,BICKER R.Fault diagnosis of industrial robot gears based on discrete wavelet transform and artificial neural network[J].Insight-Non-Destructive Testing and Condition Monitoring,2016,58(4):179-186.
[8]张铭钧,殷宝吉,刘维新,等.随机干扰下AUV推进器故障特征提取与融合[J].华中科技大学学报(自然科学版),2015,43(6):22-26.ZHANG Mingjun,YIN Baoji,LIU Weixin,et al.Fault feature extraction and fusion of AUV propeller under random interference[J].Journal of Huazhong University of Science and Technology(Natural Science Edition),2015,43(6):22-26.
[9]JABER A,BICKER R.Industrial robot backlash fault diagnosis based on discrete wavelet transform and artificial neural network[J].Journal of Quality in Maintenance Engineering,2016,11(1):21-31.
[10]ENDO M,ENDO M,KAKIZAKI T,et al.Cooperative robot system for fault diagnosis of solar panels at mega solar power plant:design of wire suspension manipulator for positioning of probe[J].Transactions of the Society of Instrument&Control Engineers,2017,53(1):22-30.
[11]ELANGOVAN K,TAMILSELVAM Y K,MOHANR E,et al.Fault diagnosis of a reconfigurable crawling-rolling robot based on support vector machines[J].Applied Sciences,2017,7(10):1025-1028.
[12]袁宪锋,宋沐民,周风余,等.多PCA模型及SVM-DS融合决策的服务机器人故障诊断[J].振动、测试与诊断,2015,35(3):434-440.YUAN Xianfeng,SONG Mumin,ZHOU Fengyu,et al.Fault diagnosis of service robots based on multiPCA model and SVM-DS fusion decision[J].Journal of Vibration,Measurement and Diagnosis,2015,35(3):434-440.
[13]武红霞,韩捷,陈宏,等.基于NOFRF裂纹转子非线性频谱分析[J].机械强度,2017,39(6):1271-1274.WU Hongxia,HAN Jie,CHEN Hong,et al.Nonlinear spectrum analysis of cracked rotor based on NO-FRF[J].Journal of Mechanical Strength,2017,39(6):1271-1274.
[14]张家良,曹建福,高峰.结合非线性频谱与贝叶斯网络的复杂装备传动系统故障诊断[J].电机与控制学报,2014,18(3):107-112.ZHANG Jialiang,CAO Jianfu,GAO Feng.Fault diagnosis of transmission system of complex equipment based on nonlinear spectrum and Bayesian network[J].Electric Machines and Control,2014,18(3):107-112.
[15]胡宇,张世英,罗雷,等.基于自适应容积卡尔曼滤波方法的涡扇发动机气路部件故障诊断[J].航空动力学报,2016,31(5):1260-1267.HU Yu,ZHANG Shiying,LUO Lei,et al.Fault diagnosis of turbofan engine pneumatic components based on adaptive volume Kalman filter[J].Journal of Aerospace Power,2016,31(5):1260-1267.
[16]张家良,曹建福,高峰,等.基于非线性频谱数据驱动的动态系统故障诊断方法[J].控制与决策,2014,29(1):168-171.ZHANG Jialiang,CAO Jianfu,GAO Feng,et al.Fault diagnosis approach of dynamic system based on data driven of non-linear spectrum[J].Control and Decision,2014,29(1):168-171.
[17]ZHU Yunpeng,LANG Ziqiang.Design of nonlinear systems in the frequency domain:an output frequency response function-based approach[J].IEEE Transactions on Control Systems Technology,2017,26(4):1358-1371.
[18]李志农,刁海洋,肖尧先.基于非线性输出频率响应函数的转子碰摩故障诊断方法研究[J].失效分析与预防,2016,11(2):67-71.LI Zhinong,DIAO Haiyang,XIAO Yaoxian.Fault diagnosis method of rub-impact in rotor system based on non-linear output frequency response function[J].Failure Analysis and Prevention,2016,11(2):67-71.
[19]张洪钺,杨萍.基于非线性状态观测器的无刷电机故障诊断[J].电机与控制学报,2006,10(1):4-8.ZHANG Hongyue,YANG Ping.Fault diagnosis of brushless motor based on nonlinear state observer[J].Electric Machines and Control,2006,10(1):4-8.
[20]阳同光,桂卫华.基于KPCA与RVM感应电机故障诊断研究[J].电机与控制学报,2016,20(9):89-95.YANG Tongguang,GUI Weihua.Research on fault diagnosis of induction motor based on KPCA and RVM[J].Electric Machines and Control,2016,20(9):89-95.
[2]MA Hongjun,YANG Guanghong.Simultaneous fault diagnosis for robot manipulators with actuator and sensor faults[J].Information Sciences,2016,366(10):12-30.
[3]陈钢,郭雯,贾庆轩,等.基于运动学模型重构的单关节故障机械臂容错路径规划[J].控制与决策,2018,33(8):1436-1442.CHEN Gang,GUO Wen,JIA Qingxuan,et al.Faulttolerant path planning of single-joint fault manipulator based on kinematics model reconstruction[J].Control and Decision,2018,33(8):1436-1442.
[4]刘冠初,熊静琪,乔林,等.四足机器人自由步态规划建模与算法实现[J].西安交通大学学报,2015,49(6):84-89.LIU Guanchu,XIONG Jingqi,QIAO Lin,et al.Modelling and algorithm realization of free gait regulation for quadruped robots[J].Journal of Xi’an Jiaotong University,2015,49(6):84-89.
[5]盖海松,徐方,贾凯,等.一种机器人关节扭矩传感器的设计与研究[J].现代制造工程,2017(5):33-37.GAI Haisong,XU Fang,JIA Kai,et al.Design and research of a robot joint torque sensor[J].Modern Manufacturing Engineering,2017(5):33-37.
[6]严浩,白瑞林,吉峰.一种改进的SCARA机器人动力学参数辨识方法[J].中国机械工程,2017,28(22):2707-2713.YAN Hao,BAI Ruilin,JI Feng.An improved dynamic parameter identification method for SCARA robots[J].China Mechanical Engineering,2017,28(22):2707-2713.
[7]JABER A,BICKER R.Fault diagnosis of industrial robot gears based on discrete wavelet transform and artificial neural network[J].Insight-Non-Destructive Testing and Condition Monitoring,2016,58(4):179-186.
[8]张铭钧,殷宝吉,刘维新,等.随机干扰下AUV推进器故障特征提取与融合[J].华中科技大学学报(自然科学版),2015,43(6):22-26.ZHANG Mingjun,YIN Baoji,LIU Weixin,et al.Fault feature extraction and fusion of AUV propeller under random interference[J].Journal of Huazhong University of Science and Technology(Natural Science Edition),2015,43(6):22-26.
[9]JABER A,BICKER R.Industrial robot backlash fault diagnosis based on discrete wavelet transform and artificial neural network[J].Journal of Quality in Maintenance Engineering,2016,11(1):21-31.
[10]ENDO M,ENDO M,KAKIZAKI T,et al.Cooperative robot system for fault diagnosis of solar panels at mega solar power plant:design of wire suspension manipulator for positioning of probe[J].Transactions of the Society of Instrument&Control Engineers,2017,53(1):22-30.
[11]ELANGOVAN K,TAMILSELVAM Y K,MOHANR E,et al.Fault diagnosis of a reconfigurable crawling-rolling robot based on support vector machines[J].Applied Sciences,2017,7(10):1025-1028.
[12]袁宪锋,宋沐民,周风余,等.多PCA模型及SVM-DS融合决策的服务机器人故障诊断[J].振动、测试与诊断,2015,35(3):434-440.YUAN Xianfeng,SONG Mumin,ZHOU Fengyu,et al.Fault diagnosis of service robots based on multiPCA model and SVM-DS fusion decision[J].Journal of Vibration,Measurement and Diagnosis,2015,35(3):434-440.
[13]武红霞,韩捷,陈宏,等.基于NOFRF裂纹转子非线性频谱分析[J].机械强度,2017,39(6):1271-1274.WU Hongxia,HAN Jie,CHEN Hong,et al.Nonlinear spectrum analysis of cracked rotor based on NO-FRF[J].Journal of Mechanical Strength,2017,39(6):1271-1274.
[14]张家良,曹建福,高峰.结合非线性频谱与贝叶斯网络的复杂装备传动系统故障诊断[J].电机与控制学报,2014,18(3):107-112.ZHANG Jialiang,CAO Jianfu,GAO Feng.Fault diagnosis of transmission system of complex equipment based on nonlinear spectrum and Bayesian network[J].Electric Machines and Control,2014,18(3):107-112.
[15]胡宇,张世英,罗雷,等.基于自适应容积卡尔曼滤波方法的涡扇发动机气路部件故障诊断[J].航空动力学报,2016,31(5):1260-1267.HU Yu,ZHANG Shiying,LUO Lei,et al.Fault diagnosis of turbofan engine pneumatic components based on adaptive volume Kalman filter[J].Journal of Aerospace Power,2016,31(5):1260-1267.
[16]张家良,曹建福,高峰,等.基于非线性频谱数据驱动的动态系统故障诊断方法[J].控制与决策,2014,29(1):168-171.ZHANG Jialiang,CAO Jianfu,GAO Feng,et al.Fault diagnosis approach of dynamic system based on data driven of non-linear spectrum[J].Control and Decision,2014,29(1):168-171.
[17]ZHU Yunpeng,LANG Ziqiang.Design of nonlinear systems in the frequency domain:an output frequency response function-based approach[J].IEEE Transactions on Control Systems Technology,2017,26(4):1358-1371.
[18]李志农,刁海洋,肖尧先.基于非线性输出频率响应函数的转子碰摩故障诊断方法研究[J].失效分析与预防,2016,11(2):67-71.LI Zhinong,DIAO Haiyang,XIAO Yaoxian.Fault diagnosis method of rub-impact in rotor system based on non-linear output frequency response function[J].Failure Analysis and Prevention,2016,11(2):67-71.
[19]张洪钺,杨萍.基于非线性状态观测器的无刷电机故障诊断[J].电机与控制学报,2006,10(1):4-8.ZHANG Hongyue,YANG Ping.Fault diagnosis of brushless motor based on nonlinear state observer[J].Electric Machines and Control,2006,10(1):4-8.
[20]阳同光,桂卫华.基于KPCA与RVM感应电机故障诊断研究[J].电机与控制学报,2016,20(9):89-95.YANG Tongguang,GUI Weihua.Research on fault diagnosis of induction motor based on KPCA and RVM[J].Electric Machines and Control,2016,20(9):89-95.