自适应MCKD和ALIF的滚动轴承早期故障诊断

Adaptive MCKD and ALIF Scrolling Early Baring Fault Diagnosis

针对滚动轴承故障信号冲击成分能量往往较低,故障特征频率难以提取以及最大相关峭度反褶积(Maximum Correlation Kurtosis Deconvolution,MCKD)降噪效果受限于滤波器L和位移数M等问题,提出了一种自适应最大相关峭度反褶积和自适应局部迭代滤波(Adaptive Local Iterative Filter,ALIF)的滚动轴承故障特征提取方法。以排列熵为标准,应用步长搜寻法确定最佳的MCKD滤波器的长度和位移数,对采集的振动信号进行降噪预处理,突出被噪声所淹没的故障冲击;然后应用ALIF算法对降噪后的信号自适应分解为一组固有模态函数(IMF)分量,利用最大峭度准则选取包含故障信息量最大的分量,即敏感分量;最后对敏感分量进行包络谱分析,提取故障特征频率。仿真和试验分析结果证明了该方法的有效性和准确性。

For the rolling bearing fault signal,the impact component energy tends to be low,the fault characteristic frequency is difficult to extract,and the Maximum correlation kurtosis deconvolution(MCKD)noise reduction effect is limited by the filter L and the displacement number M.A method for extracting fault characteristics of rolling bearings based on adaptive maximum correlation kurtosis deconvolution and Adaptive local iterative filter(ALIF).Taking the entropy as the standard,the step search method is used to determine the length and displacement of the optimal MCKD filter,and the collected vibration signal is denoised and pre-emphasized to highlight the fault impact that is overwhelmed by noise;then the ALIF algorithm is used to reduce the impact.The signal after noise is adaptively decomposed into a set of intrinsic mode function(IMF)components.The maximum kurtosis criterion is used to select the component containing the largest amount of fault information,that is,the sensitive component.Finally,the envelope spectrum analysis of the sensitive component is performed to extract the fault feature.frequency.Simulation and experimental analysis results demonstrate the effectiveness and accuracy of the proposed method.