Detection of rub-impact fault for rotor-stator systems: A novel method based on adaptive chirp mode decomposition

Abstract

For rotating machineries, rub-impact between the surfaces of the rotor and the stator during the rotation is a common and severe fault. The rub-impact fault will lead to a frequency-modulated (FM) vibration signal with a fast fluctuating instantaneous frequency (IF). It is difficult to extract the fast fluctuating IF feature due to the limited resolution of the current time-frequency (TF) methods. The recently proposed variational nonlinear chirp mode decomposition (VNCMD) method shows promising advantages in analyzing strongly FM signals. However, the VNCMD solves a joint-estimation problem which requires the prior information of the number of the signal components and may cause instability issues. In this paper, a tractable version of the VNCMD, called adaptive chirp mode decomposition (ACMD), is introduced to extract the fast fluctuating IF of the vibration signal from the rub-impact rotor. The ACMD employs a greedy algorithm to catch each signal component individually. Moreover, using the estimated instantaneous amplitude and the IF by ACMD, we obtain a high-resolution adaptive TF spectrum which can clearly represent the rub-impact feature of the vibration signal. The effectiveness of the fault detection method based on ACMD is demonstrated by dynamic simulations at first. Then, the method is applied to vibration signals of a heavy oil catalytic cracking machine set indicating its usefulness in early fault detection and multi-feature extraction.