基于LCD-Hilbert谱奇异值和QRVPMCD的滚动轴承故障诊断方法  

Rolling bearing fault diagnosis method based on Hilbert spectrum singular values and QRVPMCD

作　　者：杨宇[1] 何知义 潘海洋[1] 程军圣[1] 

机构地区：[1]湖南大学汽车车身先进设计制造国家重点实验室,长沙410082

出　　处：《振动与冲击》2015年第7期121-126,共6页Journal of Vibration and Shock

基　　金：国家自然科学基金(51175158;51375152);湖南省自然科学基金(11JJ2026)

摘　　要：针对多变量预测模型的模式识别(Variable Predictive Model Based Class Discriminate,VPMCD)方法在参数估计中存在的缺陷,采用分位数回归(Quantile Regression,QR)代替原方法中的最小二乘法进行参数估计,克服最小二乘回归中强假设、易受异常值影响等问题,以此提高模式识别的精度。因此,提出了基于分位数回归的多变量预测模型模式识别方法(Quantile Regression-Variable Predictive Mode Based Cass Discriminate,QRVPMCD)。采用局部特征尺度分解(Local Characteristic-Scale Decomposition,LCD)方法对滚动轴承振动信号进行分解得到若干个单分量信号,提取单分量信号的Hilbert谱奇异值组成故障特征向量,并以此作为QRVPMCD的输入进行滚动轴承故障诊断。对不同工作状态和故障类型下的滚动轴承振动信号进行了分析,结果表明了该方法的有效性。Targeting the defects in the parameter estimation of VPMCD （variable predictive model-based class discriminate）, Quantile Regression （QR） was used for parameter estimation instead of least-square approach in the original method. The questions such as strong assumptions and easiness of being affected by the outliers in the ordinaryleast-square regression could be overcome by QR so as to improve the accuracy of pattern recognition. Therefore, the quantile regression-variable predictive mode based on class discriminate （QRVPMCD） was proposed. The local characteristic-scale decomposition （LCD） was used to decompose the rolling bearing vibration signal into several mono- component signals, and then the Hilbert spectrum singular values were extracted from the mono-component signals a fault feature vector, which was then used as input of QRVPMCD for roiling bearing fault diagnosis. The analysis to form resultsunder different working conditions and different kinds of failures of roller bearings demonstrate the effectiveness of the proposed method.

关 键 词：QRVPMCD LCD Hilbert谱奇异值 滚动轴承 故障诊断 

分 类 号：TH113[机械工程—机械设计及理论]