Application of Ensemble Empirical Mode Decomposition and Correlated Kurtosis to Rolling Element Bearing Fault Diagnosis

The vibration signals created by rolling bearing are very complicated,especially for those bearing applied in ship machinery equipment. Normally the fault signal is easily covered by background noise,so the traditional methods can't effectively extract the weak fault information. In order to solve this problem,the EEMD and correlated kurtosis is applied to rolling bearing fault diagnosis,and a feature extraction method is proposed. According to the method,a group of component signals are obtained by EEMD,then the component signal with maximal correlated kurtosis is selected,and the faint fault information is extracted. Through simulation and experiment,the effectiveness of the new method is demonstrated.