Abstract: Feature extraction and fault diagnosis method are particularly important for diagnosing the faults of aero-engine rotor quickly and accurately.Aiming at the non-stationary and nonlinear characteristics of aero-engine's vibration signals,the wavelet transform and Hilbert-Huang Transform (HHT) method were used to extract the three feature vectors that were the effective value of signal,the marginal spectrum centroid and the power spectral centroid of maximum energy level which was obtained with wavelet transform.The two-stage adaptive wavecluster was applied to diagnose the mixed fault of aero-engine rotor.The process of two-stage adaptive wavecluster was that: the large grid cell was used to quantify the data space,and the area of clustering was found to achieve presorting clustering of data;the sub cluster region was subdivided adaptively again,and wavecluster was implemented in this region;compared with the fault sample,the fault type of cluster was identified.The results showed that the proposed method could achieve fault classification and recognition quickly and accurately.Especially for the mixed multi-type data with non-uniform density,the diagnostic accuracy was significantly higher than the traditional wavecluster.

Key words: fault diagnosis, wavecluster, feature extraction, two-stage adaptive

摘要： 有效特征向量的提取与故障诊断方法是实现航空发动机转子故障快速正确诊断的关键。首先根据航空发动机转子振动信号的非平稳及非线性的特点,应用小波变换和Hilbert-Huang变换方法提取振动信号的有效值、边际谱重心和小波变换最大能量层的功率谱重心三个特征向量,然后应用双阶自适应小波聚类方法对航空发动机转子进行多类型混合故障诊断。结果表明双阶自适应小波聚类方法能快速准确地实现故障分类与识别,尤其对于密度分布不均匀的多类型混合数据,诊断精度显著高于传统的小波聚类方法。

关键词: 故障诊断, 小波聚类, 特征提取, 双阶自适应