摘要
轴承故障分析在滚动传动系统中一直是研究的热点,传统的轴承故障诊断方法往往建立在苛刻的约束条件之上,如检测信号为单一的故障信号成分、既定的混合系统保持不变或者模型建立在无噪声的环境等。针对这些局限,结合了独立成分分析(Independent Component Analysis,ICA)方法,提出了一种基于特征矩阵联合相似对角化及经验模态分解(Joint Approximative Diagonalization of Eigen matrix-Empirical Mode Decomposition,JADE-EMD)的多故障动态盲分析技术。该方法的基本思想是基于多输入多输出的动态混合模型,利用四阶统计量对随机噪声的盲辨识特性,将滚动轴承正常工作时的平稳随机噪声看成一类常规的信号输入。接着通过动态的盲源分离技术将传感器接收到的混合信号分解成相互独立的成分,最后对分离的故障信号进行EMD分解,并得到多个基本模式分量函数(Intrinsic Mode Function,IMF)的分布结果。仿真研究表明,该方法可以对带有故障的滚动轴承进行有效的诊断,特别是在背景噪声较强的多轴承传动系统中,能够有效避免多种故障信号之间的相互干扰,相对于传统的单一直接检测方法而言,可以进一步提高对故障轴承分析的准确性。
Bearing fault analysis has been a research focus in rolling transmission system.However,the traditional bearing fault diagnosis technology is usually based on strict constraints,such as the detection signal is a single fault signal component,the established hybrid system remains unchanged,and the model is established in noise free situation.Aiming at the limitation of this problem,combined with the independent component analysis(ICA)method,this study proposes a multi fault dynamic blind analysis method based on joint approximate diagonalization of eigenmatrix empirical mode decision(JADE-EMD).The basic idea of this method is based on the dynamic transmission system with multi input and multi output.Because of the blind identification characteristics for random noise with fourth-order statistics,the stationary random noise of rolling bearing in normal operation works as a kind of conventional signal input.Then,the mixed signals received by the sensor are decomposed into independent components by dynamic blind source separation technology.Finally,the separated fault signals are decomposed by EMD,and the distribution results of several basic mode component functions(IMF)are obtained.Simulation results show that the method can effectively diagnose the rolling bearing with faults.Especially in the multi bearing drive system,it can effectively avoid the mutual interference between various fault signals.Compared with the traditional single direct detection method,it can further improve the accuracy of fault bearing analysis.
作者
冯平兴
张洪波
Feng Pingxing;Zhang Hongbo(School of Network and Communication Engineering,Chengdu Technological University,Chengdu 611731,China;School of Communication and Information Engineering,Chengdu University of Information Technology,Chengdu 610225,China)
出处
《电子技术应用》
2021年第6期71-76,共6页
Application of Electronic Technique