2理论背景-仪器仪表学报.doc

基于SK-NLM包络的滚动轴承故障冲击特征增强 熊国良，胡俊锋，陈 慧，张 龙 (华东交通大学机电工程学院南昌330013) 摘要：non-local Means，NLM）是活跃于图像信号处理领域的一种新方法，因其良好的去噪特性，近几年来在滚动轴承故障诊断领域也开始获得应用。NLM利用样本点邻域窗口包含的局部结构为基本单元，通过对相似成分加权运算后取其平均值以达到抑制噪声干扰、突出故障冲击特征的目的。但对于强噪声条件下的低信噪比信号而言，NLM滤波效果并不理想。提出一种结合谱峭度（spectral kKurtosis，SK）和NLM权重包络谱的故障诊断方法，首先对原始信号进行SK分析得到最优中心频率及带宽构成最优滤波器，初步消除环境干扰及测量噪声；其次对NLM算法进行改进，不再以滤波信号为分析对象，而是直接利用NLM加权运算得到的信号样本点权值分布曲线作为预处理信号的包络信号，从权重角度使故障冲击得到二次增强，消除SK带通滤波器的带内噪声；最后对权值分布曲线进行包络谱分析，进而得到诊断结果。通过仿真信号、实验室信号及工程应用信号分析对所提方法进行了验证，并与最小熵解卷积（minimum entropy deconvolutionMED） 关键词：滚动轴承；谱峭度；非局部均值算法；加权运算；故障诊断 中图分类号：TH165.3TN911.7 文献标识码：A Feature Enhancement for Rolling Bearings by Spectral Kurtosis and Non-Local Means XIONG Guoliang，HU Junfeng，CHEN Hui，ZHANG Long (School of Mechatronics Vehicle Engineering, East China Jiaotong University, Nanchang 330013, China) Abstract: The nonlocal means (NLM) is a new method which attracts significant attention in the field of image processing as it overcomes the limitations of the neighborhood filter effectively. To date, NLM has been extended to rolling bearing fault diagnosis due to its ability to reduce the background. Unfortunately, the diagnosis result of NLM is not perfect in the case of low SNR. Aiming at such problem, a hybrid approach using spectral kurtosis (SK) and Non-Local Means has been proposed to detect rolling element bearing faults. The approach consists of the following three main steps. Firstly, the optimal center frequency and bandwidth is determined by spectral kurtosis analysis for the optimal filter which has the ability to eliminate environmental disturbance and measurement noise. And then the cyclic impact feature is further enhanced in the form of weight distribution curve by the weighted operation of NLM. Finally, the characteristic fault frequencies could be readily detected through the spectrum of weight series. The effectiveness and merits of the proposed approach were verified by numerical simulation and experimental data of rolling element bearing wit