WAVELET ANALYSIS FOR SPINDLE VIBRATION SIGNAL PROCESSING UNDER NON-UNIFORM PRELOAD

摘要

Skewing or dislocation installation of a bearing in a spindle will be subjected to uneven loads caused by machining error, assembly error and thermal deformation. Bending moment will be induced by the uneven force around the center of bearing. Non-uniform preload methodology is proposed in this paper aiming to stimulate the skewing installation in a spindle. Wavelet threshold technique for noise reduction and wavelet package method for feature extraction were introduced in order to identify the bearing posture through analyzing the vibration signal of the bearing. To validate the proposed method, the raw vibration signals are reconstructed, the eigenvector is structured and normalization processing is done between uniform preload and non-uniform preload conditions, respectively. The analysis results shows that the wavelet packet threshold de-noising technology can effectively eliminate the noise of the spindle vibration signal. The bearing posture identification and bearing preload identification are realized through the change of frequency energy.