Study on Feature Extraction of High Speed Precision Electric Machine Vibration Signal

摘要

Vibration signals usually contain running condition and fault information of rolling mechanical equipment. In the paper, firstly, a vibration test scheme of high speed precision electric machine is designed; time domain average method is used to filter the periodic noise and random noise of the sampling vibration signals. The result shows that the signal to noise ratio is increased. Then the denosing vibration signal is decomposed by means of wavelet packet and the reconstructed signal energy of every frequency segment is calculated. The study identifies that the reconstructed signal of every frequency segment contains corresponding frequency, and the energy can be used as the vibration signals' eigenvector to estimate the running state of the electric machine.