Fault feature identification based on fast analytical modal decomposition method for bearing diagnosis

Abstract The analytical mode decomposition (AMD) method has been employed in structural mode identification, additional investigation and study are required in bearing and gear fault detection. This paper proposed a fast analytical mode decomposition (FAMD) technology to improve the flexibility and accuracy of AMD in the process of bearing fault diagnosis. 1/3 binary tree structure is used to create a multi-partition framework that can estimate modes and binary frequency groups rapidly and effectively. Mode extraction methods based on analytical mode decomposition theory can produce more precise filtering results. Finally, an enhanced hybrid index based on envelope information is intended to effectively detect periodic pulses in each component and diagnose bearing faults. Simulation signals and bearing failure data from Case Western Reserve University verify the effectiveness of the proposed method.