基于样本分位数排列熵的故障诊断方法

Fault diagnosis method based on sample quantile permutation entropy

针对滚动轴承故障特征不明显、不易于进行特征提取等问题,提出了一种新的衡量时间序列复杂程度的方法-样本分位数排列熵(Sample Quantile Permutation Entropy,SQPE),并将其应用于滚动轴承故障振动信号的特征提取。通过对振动信号进行样本分位数排列熵计算,有效分离出不同振动信号的故障特征;将熵值组成特征向量,构建分类器并实现对滚动轴承的故障诊断。将提出的方法应用于试验数据分析,结果表明:样本分位数排列熵能够有效提取滚动轴承的故障特征,并在熵值计算的过程中,避免了嵌入维数选取的过程,有效提高了熵值计算的自适应性,扩大了其应用范围。

Aiming at problems of rolling bearings’obscure fault features and difficult to extract features,a new method to measure the complexity of time series,i.e.,the sample quantile permutation entropy(SQPE)method was proposed and applied in feature extraction of rolling bearing fault vibration signals.By calculating sample quantile permutation entropies of fault vibration signals,fault features of different vibration signals were effectively separated.Then,the obtained entropy values were formed into feature vectors to construct a classifier,and realize fault diagnosis of rolling bearings.The proposed method was applied to analyze test data.The results showed that SQPE method can be used to effectively extract fault features of rolling bearings;the process of calculating entropy values avoids the selecting process of embedded dimension to effectively improve the self-adaptability of entropy value calculation,and expand its application range.