Adaptive direct fast iterative filtering based rolling bearing fault diagnosis

Direct fast iterative filtering (DFIF) is a recently proposed nonlinear and nonstationary signal analysis method. Aiming at the problems that the DFIF method needs to preset the adjustment parameter of filtering interval, and lacks adaptability in the process of iterative calculation, an adaptive direct fast iterative filtering (ADFIF) method was proposed based on the instantaneous frequency fluctuation energy difference criterion, which could adaptively determine the optimal filter interval adjustment parameters in the iterative screening process of each layer in the outer loop of the DFIF algorithm. The ADFIF method can adaptively decompose a given nonlinear and non-stationary signal into the sum of several approximately narrow band signals whose instantaneous frequency has physical significance and besides, a trend term. Through the simulated and measured signal analysis of rolling bearings, the results of the proposed ADFIF method were compared with those of the original DFIF, the adaptive local iterative filtering, the variational mode decomposition, and the empirical mode decomposition. The results show that the proposed ADFF method has certain advantages in suppressing mode mixing and improving anti-noise performance, and can extract more fault characteristic information of rolling bearings. © 2023 Chinese Vibration Engineering Society. All rights reserved.