Rolling Bearing’s Breakdown Feature Extraction Technology Based On Ensemble Empirical Mode Decomposition

Rotating machinery is the key of the mechanical fault diagnosis, and a number of the rotating machinery faults are relative to rolling bearing. When rolling bearing fails, the running state has great influence on the whole machine. So the condition monitoring and fault diagnosis of rolling bearing has important practical significance.This paper mainly researches bearing fault diagnosis based on the vibration signal. When rolling bearing breaks down, periodic impulse signal will be produced. Then fault characteristic values are extracted from the vibration signal to judge rolling bearing faultFist of all, rolling bearing is preliminarily diagnosed by amplitude domain diagnosis to determine whether there is fault. And then, envelope analysis usually demodulates the collected signal. Envelope signal is gained, which is time domain signal, and then spectrum is obtained through the Fourier transform (FFT). The fault characteristic frequency can directly be extracted from the spectrum. But measured actual signals tend to contain bigger noise, so fault characteristic frequency can’t effectively be demodulated by envelope analysis. Therefore, we need to process the collected signal. EMD is time-frequency processing method, which deals with nonlinear and non-stationary signal. But there is mode mixing in EMD, ensemble empirical mode decomposition (EEMD) is proposed by Huang to suppress mode mixing, which is a noise assisted data analysis method. EEMD can restore the nature of the signal and significantly improve the EMD.In the large background noise, the fault was demodulated from the measured vibration signals by envelope analysis and vibration signals of rotating machinery have the non-stationary and non-linear characteristics at the same time, a new method is proposed based on the ensemble empirical mode decomposition and envelope analysis for the rolling bearing fault diagnosis. Firstly, the vibration signals are decomposed into a finite number of intrinsic mode functions (IMFs) and one residue by EEMD. Secondly, the fault is detected from the IMFs by envelope analysis. In order to verify the viability of this method through the simulation signal and the actual vibration signals of outer race and inner race of rolling bearing, it is showed that this method can effectively identify rolling bearing fault.