Compound faults diagnosis based on customized balanced multiwavelets and adaptive maximum correlated kurtosis deconvolution

Due to the compound faults with different type fault coupled together and the nonobvious periodic impulses contaminated by strong noise, it is challenging to extract the fault characteristics from the rotating machinery. To overcome the limitations of maximum correlated kurtosis deconvolution (MCKD) and multiwavelets, a method combining customized balanced multiwavelets and adaptive MCKD is proposed for rotating mechanical compound faults diagnosis. First, the raw vibration signal is denoised by the customized balanced multiwavelets. Second, adaptive MCKD is utilized to decoupled the fault information from the denoised signal. Finally, the major fault characteristic frequency is extracted by Hilbert spectrum analysis. The feasibility and effectiveness of the method are demonstrated by the simulation signal and the experimental data on aero engine rotor experimental rig with compound faults combined by three different faults of rubbing fault, shaft misalignment fault and unbalance fault. (C) 2019 Elsevier Ltd. All rights reserved.