基于双谱的模糊聚类滚动轴承故障诊断

A Diagnosis Study of Rolling Bearing Based on Bispectrum and Fuzzy Clustering

为了能准确可靠地判别出机械设备中滚动轴承的故障类型,基于双谱能表征随机信号偏离高斯噪声程度的特性,提出一种模糊聚类的滚动轴承故障诊断方法.基于滚动轴承运行过程中监测到的振动信号包含高斯噪声且具有非高斯、非线性特点,该方法在分析高阶谱理论的基础上,对双谱估计特征值进行阈值化处理,构成二值特征图并构造模板,通过测试样本与目标模板之间距离大小来进行不同类别的判断,实现对滚动轴承故障的诊断.经实例验证,该方法有效实现了对滚动轴承故障的分类判别,并提高了判别的可靠性和准确性.

In order to judge the fault type of machinery and equipment accurately and reliably , a diagnosis method based on bispectrum and fuzzy clustering for rolling bearingwas proposed with the help of fuzzy clustering representing the deflected level. Because vibration signals monitored in the rolling operation contained Gaussian noise , which also share a characteristis of non-Gaussian and nonlinear, this proposal, on the basis of analyzing he higher-order spectrum, eigenvalued threshold processing, constited 0-1 values characteristic graph, established template, and made different types of judgments by gap bettwne sample template and target template , and then obtained fault diagonosis of rolling bearing. Then by testing the distance between the template and the target sample sizes for different types of judgment, it is concluded that this method can present different judgments acoordingly, and increase the reliablity and accuracy.