New fault diagnosis approaches for detecting the bearing slight degradation

In this paper, two new methods for detecting the bearing’s degradation starting points are presented based on the vibration signal analysis. In the first method, a new feature extraction technique is suggested based on the envelope harmonic-to-noise ratio (EHNR) and the fast ensemble empirical mode decomposition (FEEMD). Each vibration signal is decomposed into its intrinsic mode functions (IMFs) using the FEEMD algorithm. Also, a novel technique has been introduced based on the autocorrelation function (ACF) of the original signal and its IMFs for selecting the most appropriate IMF and eliminate the noisy components. Then, the EHNR of the most sensitive IMF is computed for detecting the early degradation of bearing. In the second method, a new adaptive feature is defined using the ACF of the raw signal and the energy-entropy vector. At first, a novel indicator called the periodicity intensity factor (PIF) is introduced using the energy of the ACF of the raw signal and its maximum points. In the next step, the energy-entropy variations of the PIF factor are investigated for recognizing the fault starting point in bearings. In this work, the vibration signals of the run-to-failure experiment are used to appraise the presented techniques. The results indicate that the proposed approaches are able to detect the exact moment of the defect occurrence. Also, comparing the results of this paper with other techniques presented recently indicates the superiority of the proposed approaches.