LAUSR

As a common fault of the rotor system, the rub-impact between a rotor and a stator causes the rotor system to exhibit strong nonlinearities. Therefore, the vibration signals obtained from a rotor system with the rubbing fault contain weak high-order nonlinear components, which are difficult to be characterized by traditional linear methods. The nonlinear output frequency response functions (NOFRFs) are an effective approach for analyzing nonlinear systems in the frequency domain, which has been applied by many researchers in the fault diagnosis of nonlinear systems. In the present study, the concept of Kullback–Leibler (KL) divergence is introduced to NOFRFs to develop a novel method named the improved NOFRFs weighted contribution rate based on KL divergence (NWKL). And it is employed to detect the rotor system with a rub-impact fault. As a new index extracted by NWKL, the improved NOFRFs optimal weighted contribution rate based on KL divergence (KLRm) is proposed to evaluate the severity of rub-impact faults in rotor systems. The simulation and experimental results demonstrate high sensitivity and strong feature dynamic stability of the newly proposed index to rotor rub-impact. A new evaluation method named the feature dynamic stability analysis is proposed to explore the influence of the change in the friction coefficient and rub-impact stiffness on KLRm. The methods in this paper show the important significance and potential application values in the detection and evaluation of rotor rub-impact.