Rotor Imbalance Recognition of Electric Spindle Based on Wavelet Packet and Random Forest

Abstract

Aiming at the problem that it is difficult to identify and classify the rotor imbalance of the electric spindle, a dynamic balance test bench of the electric spindle is built, and the rotor imbalance experiment at different speeds is performed, and vibration signals are collected. 1. The wavelet packet method is adopted to denoise the vibration signal. 2. The four characteristic parameters of amplitude, variance, standard deviation, and mean square error are selected by tSEN cluster analysis to combine into the rotor imbalance state evaluation model. 3. The combined evaluation model is input into the chosen random forest for training and identification. The results show that the rotor imbalance evaluation model established in this paper can accurately and effectively identify different types of rotor imbalance. It is better than time-domain feature model, frequency-domain feature model and wavelet packet feature model in terms of time-consuming and accuracy.