Abstract: The period recognition ability of the existing rolling bearing fault diagnosis methods (such as wavelet transform and ensemble empirical mode decomposition) is not stable. We propose adaptive periodic mode decomposition (APMD) method with good capability of extracting periodic components. However, the maximum likelihood estimation method used in the APMD often fails to estimate the period under strong noise background, which results in the unstable performance of period extraction under strong noise background. Therefore, we define an adaptive frequency weighted energy operator (AFWEO) and use it to enhance period impulses. Then, we propose a new period estimation strategy to improve the accuracy of the period estimation. On this basis, we propose the enhanced Ramanujan mode decomposition (ERMD) method. Emulational and experimental signal analysis results of rolling bearings show that the new period estimation strategy is still effective under strong background noise, and ERMD is an effective method for rolling bearing fault diagnosis because of its excellent ability of identifying and extracting periodic components.

Key words: enhanced ramanujan mode decomposition, period estimation, adaptive frequency weighted energy operator, rolling bearing, fault diagnosis