A Novel Strategy for Signal Denoising Using Two-Layer SVD and Its Application to Rub-Impact Fault Diagnosis of Aeroengine

Abstract

The rub-impact fault, which occurs in the aeroengine, can cause frequency modulation (FM) in its vibration signal. Oscillated instantaneous frequency (IF) is a phenomenon of such a modulation mode, which carries critical fault information from the aeroengine. In this paper, by using Variational Mode Decomposition (VMD), the vibration signal can be decomposed into several signal components. Two-layer Singular Value Decomposition (SVD) is utilized to reduce the signal noise of each signal component and instantaneous frequency of each signal component. The intra–wave FM feature of the signal are presented through the Hilbert transform (HT) analysis. The proposed method is validated with numerical simulation and then demonstrated on aeroengine data with the rub-impact fault. The proposed method is suitable for the analysis of modulated signals and is a valid tool for detecting the rub-impact faults in aeroengines.