Improved EEMD Algorithm and Its Application to Fault Diagnosis for Rolling Bearing of Wind Turbine

This paper discusses an improved method for the ensemble empirical mode decomposition (EEMD). To improve the decomposition effect, the mutual information is used to the algorithm. In order to deal with the mode mixing problem, two white noises with opposite signs are added to the original signal. Before the average calculating operation for the intrinsic mode functions (IMFs), the mutual information operation is processed, and the false IMFs are eliminated according to the mutual information threshold. To illustrate the effect of the improved algorithm, the simulated analysis is given in detail, and the analysis results show that the improved algorithm has better decomposition effect than EEMD. As example of application, the improved algorithm has been used to fault diagnosis for the rolling bearing of wind turbine, and the decomposition results are given. The envelope spectrums for different algorithms are presented, and detailed analysis is processed. By comparing the envelope spectrums with the other algorithms, the improved algorithm is proved having the best performance.