An improved NOFRFs-based fault feature extraction method and its application to quantitative diagnosis in rotor rub-impact

Abstract In this paper, the Jeffcott rotor model has been set up to extract and describe the nonlinear characteristics caused by the local rub-impact. During a mechanical fault diagnosis, the detection methods frequently lead to misjudgment only based on the output behaviors. Hence, nonlinear output frequency response functions (NOFRFs) have gradually developed. Like the linear transfer function, NOFRFs can reflect the relationship between the spectrums and the nonlinear system parameters. However, in the fault diagnosis of the rotor system, the current fault feature extraction methods based on NOFRFs can describe the damage severity tend qualitatively, but diagnostic results cannot be guaranteed by only analyzing the vibration response under a single rotational speed. In the present study, a definite amplification coefficient αn is clarified to keep the same magnification ratio of the high-order NOFRFs. It achieves the purpose that describes the amplified fault characteristics quantitatively. Furthermore, a novel fault diagnosis criterion—weighted contribution ergodic index (TR(n)), is proposed to consider system responses in a wide range of harmonic excitations, so a more comprehensive system characteristic can be obtained to make the diagnosis results more reliable. Simulation results have demonstrated that the new fault diagnosis method can enhance the sensitivity and stability to identify the rub-impact fault. Simultaneously, the experiments based on the rotor bench have verified the superiority of the new method in practical application.