关键词: 改进VPMCD, 局部特征尺度分解, 主成分估计, 滚动轴承, 故障诊断

Abstract:

According to the drawbacks in the parameter estimation process of a new pattern recognition approach—VPMCD, the VPMCD approach was improved, its parameter estimation approach,least square method, was replaced by principal component estimation. The principal component estimation can eliminate the influences of the multiple linear correlations among the predictive variables. As a result, it can get more stable model parameters and improve the precision of pattern recognition. LCD approach was first adopted herein to decompose the roller bearing vibration
signals into several mono-component signals. Then the approximate entropies were abstracted from the mono-component signals and formed into fault feature vector which will act as input in the improved VPMCD. Finally, the improved VPMCD was used as a classifier to distinguish the different working conditions and failures of roller bearing. The analysis results from vibration signals of the four roller bearing states i.e. normal, outer faults, inner faults and roller faults demonstrate the effectiveness of the proposed method.

Key words: improved variable predictive model based class discriminate(VPMCD);local characteristic-scale decomposition(LCD), principal component estimation;roller bearing;fault diagnosis