Research on test bench bearing fault diagnosis of improved EEMD based on improved adaptive resonance technology

Ensemble empirical mode decomposition (EEMD) is an adaptive signal decomposition method. The selection of the optimal intrinsic mode function (IMF) and the enhancement of the de-noising ability of EEMD have always been the problems that researchers are attempting to solve. The paper proposes an improved EEMD based on the improved adaptive resonance technology (IART) to resolve the above problems. The main work of this paper is described as: At first, the IART theory is summarized based on the traditional resonance technology. Secondly, a novel method to select the optimal IMF(s) based on the resonance frequency (RF) of IART is put forward. Thirdly, for the determination of the parameters of the band-pass filter, an optimization method IART-based is presented, and the parameters are optimized by the RF and the principle of the maximum of envelope kurtosis (MEK) to solve the problem that the center frequency and the bandwidth of the traditional resonance technology determined by experience. Finally, the IART is used as a supplement to EEMD to enhance its de-noising ability. Experimental results on simulated signal and vibration signals measured from rolling bearings have revealed that the improved EEMD can obtain a more satisfactory effect than other commonly used methods.