Two-Step Adaptive Chirp Mode Decomposition for Time-Varying Bearing Fault Diagnosis

Fault diagnosis of rolling bearing has always been the focus of research in the engineering field. Due to the nonstationary conditions, this work has suffered huge challenges. The adaptive chirp mode decomposition (ACMD) is a signal decomposition method developed recently. By using a recursive framework embedded with an iterative algorithm, it can decompose signals one by one. Although this scheme has strong adaptability, the accuracy of reconstructed signals is very rough. Moreover, the estimated components will even deviate from the true results as the iteration progresses when two components are very close. In order to solve the above problems, a novel two-step ACMD is proposed in this article. Specifically, the recursive scheme without an iterative algorithm is only used to decompose signals first, which can determine the number of components. Then, the joint-estimation framework using an iterative algorithm is used to achieve high-resolution component reconstruction. The main idea of this two-step method is to let different schemes do what they do best. Subsequently, we proposed a new fault diagnosis method for rolling bearing under variable speeds based on the proposed two-step ACMD. Finally, both simulation and actual bearing diagnosis cases verify the potential of the proposed method and successfully diagnosed different kinds of rolling hearing faults.

Automated summary

The study proposes a two-step ACMD decomposition method, which first uses a recursive scheme to determine the number of components and then uses an iterative algorithm to achieve high-resolution component reconstruction. The method focuses on fault diagnosis of rolling bearing under variable speeds.