滚动轴承故障特征自适应增强的相关峭度联合降噪方法

针对滚动轴承早期故障阶段冲击特征微弱且存在信号传输路径、强背景噪声以及高幅值偶然性冲击的干扰,导致故障特征成分难以提取以及大多滚动轴承故障复合诊断方法处理步骤采用的诊断优化指标不一致等问题,提出了一种滚动轴承故障特征自适应增强的相关峭度(CK)联合降噪方法。首先以CK为指标优化最大相关峭度解卷积(Maximum Correlated Kurtosis Deconvolution,MCKD)参数T后对原始信号进行预处理,以消除信号传输路径的影响。进一步以相关峭度作为优化指标结合粒子群优化算法(Particle swarm optimization,PSO)优化变分模态分解(Variational mode decomposition,VMD)参数,将预处理后信号分解成若干个窄带本征模态分量(IMF),再以CK最大为准则选择故障信息含量丰富的本征模态分量,达到去噪的目的。最后结合1.5维能量谱增强信号瞬时冲击特征的优点,对滚动轴承故障类型进行诊断。文中所提主要的创新点是前后处理优化指标的一致性。仿真信号与实测信号分析结果表明,该方法可以有效剔除噪声、高幅值偶然性冲击对故障振动信号的影响,能够清晰地提取出滚动轴承早期故障信号中的故障特征频率成分,实现滚动轴承早期故障的有效判别。

Adaptive Method of Correlation Kurtosis and De-noise for Fault Characteristics of Rolling Bearings

In the early stage of rolling bearing fault,the impact characteristics are weak and there are interference of signal transmission path,strong background noise and high-amplitude accidental impact,which makes it difficult to extract fault feature components and inconsistent diagnostic optimization indicators used in the processing steps of most rolling bearing fault diagnosis methods.In this paper,a joint noise reduction method based on correlation kurtosis(CK)with adaptive enhancement of rolling bearing fault features is proposed.Firstly,CK is used to optimize the parameter T of Maximum Correlated Kurtosis Deconvolution(MCKD)to preprocess the original signal to eliminate the influence of signal transmission path.Furthermore,the correlation kurtosis is used as the optimization index and the particle swarm optimization(PSO)is used to optimize the parameters of variational mode decomposition(VMD).The pre-processed signal is decomposed into several narrowband intrinsic mode components(IMFs).Then the intrinsic mode components with abundant fault information are selected according to the maximum CK criterion to achieve denoising.Finally,the fault types of rolling bearings are diagnosed by combining the advantages of 1.5-dimensional energy spectrum enhancement signal instantaneous impact characteristics.The main innovation of this method is consistency of optimization indexes before and after processing.The analysis results of simulation signal and measured signal show that this method can effectively eliminate the influence of noise and high amplitude accidental impact on fault vibration signal,and can distinctly extract the fault characteristic frequency components in the early fault signal of rolling bearing,so as to realize the effective identification of early fault of rolling bearing.