基于EEMD-SVD与时域分析的马田系统轴承故障诊断  

Bearing fault diagnosis for Mahalanobis-Taguchi system based on EEMD-SVD and time-domain analysis

作　　者：剡昌锋[1] 王伟[1] 王慧滨 朱涛 吴黎晓[1] YAN Chang-feng;WANG Wei;WANG Hui-bin;ZHU Tao;WU Li-xiao(College of Mechano-Electronic Engineering, Lanzhou Univ. of Tech., Lanzhou 730050, China;92822 Troops, Zhangzhou 363000, China;69224 Troops, Akesu 842000, China)

机构地区：[1]兰州理工大学机电工程学院,甘肃兰州730050 [2]92822部队,福建漳州363000 [3]69224部队,新疆阿克苏842000

出　　处：《兰州理工大学学报》2019年第3期39-45,共7页Journal of Lanzhou University of Technology

基　　金：国家自然科学基金(51165018,51765034)

摘　　要：针对轴承初始故障发生的时间点以及退化趋势,提出了基于总体平均经验模式分解和奇异值分解方法(EEMD-SVD)与时域分析的马田系统故障诊断方法.该方法通过提取振动信号时域和时频特征,构建不同特征参数下的基准空间并利用正交表对特征参数进行降维和优化,最终融合成单一特征参数马氏距离.分别用马氏距离监测轴承运行状态,判断初始故障发生的时刻以及演化趋势,并依据马氏距离对轴承故障发展的过程进行了划分.该方法有效地提取了振动信号时频特征并优化了马田系统基准空间,更加准确地识别了轴承初始故障发生的时间点以及更加合理地划分了轴承的退化过程.通过两组滚动轴承加速寿命试验,验证了该方法的有效性和合理性.Aimed at the occurrence time point of the initial bearing fault and its degradation trend, a diagnosis method of bearing fault in Mahalanobis-Taguchi system is presented based on the entire average experience mode decomposition-singular value decomposition ( EEMD-SVD) and the time-domain analysis. In the method presented, the time-domain and time-frequency characteristics of the vibration signal is extracted to construct a datum space for different characteristic parameters an orthogonal list is employed to conduct dimensional reduction and optimization of the characteristic parameter, merging them eventually into a single characteristic parameter Mahalanobis distance. This Mahalanobis distance is used respectively to monitor the running state of the bearing, judge the occurrence moment and evolution trend of initial failure, and divide the development of the fault into stages. This method is effective to extract the time-frequency characteristics of the vibration signal and optimize the datum space of Mahalanobis-Taguchi system, and more accurate to identity the occurrence time point of the fault and divide rationally the bearing degradation process. The validity and rationality of the proposed method are verified by accelerative life test of two sets of rolling bearings.

关 键 词：初始故障 总体平均经验模式分解和奇异值分解方法 时频特征 马田系统 状态监测 故障阶段 

分 类 号：TH133[机械工程—机械制造及自动化]