Abstract
During its operation, a rotor system can be exposed to multiple faults, such as rub-impact, misalignment, cracks and unbalancing. When a crack fault occurs on the rotor shaft, the vibration response signals contain some nonlinear components that are considerably tougher to be extracted through some linear diagnosis methods. By combining the Nonlinear Output Frequency Response Functions weighted contribution rate (WNOFRFs) and Kullback–Leibler (KL) divergence, a novel fault diagnosis method of improved WNOFRFs is proposed. In this method, an index, improved optimal WNOFRFs (IOW), is defined to represent the nonlinearity of the faulty rotor system. This method has been tested through the finite element model of a cracked rotor system and then verified experimentally at the shaft crack detection test bench. The results from the simulation and experiment verified that the proposed method is applicable and effective for cracked rotor systems. The IOW indicator shows high sensitivity to crack faults and can comprehensively represent the nonlinear properties of the system. It can also quantitatively detect the crack fault, and the relationship between the values of IOW and the relative depth of the crack is approximately positively proportional. The proposed method can precisely and quantitatively diagnose crack faults in a rotor system.
Highlights
Rotor systems can be considered the most common and critical mechanical components in rotating machinery
On the basis of the associated index Fe proposed by Peng [27], Liu [30,31] introduced the weighting coefficient nρ and proposed the feature extraction method known as the NOFRFs weighted contribution rate (WNOFRFs)
A new fault diagnosis method known as improved WNOFRFs was developed by combining the WNOFRFs and KL divergence
Summary
Rotor systems can be considered the most common and critical mechanical components in rotating machinery. Many studies have focused on fault diagnosis methods for cracked rotor systems through vibration signals. By introducing the evidence theory to the NOFRFs, Cao et al [25] put forward a fault diagnosis method for complicated nonlinear systems by combining the NOFRFs and evidence theory. Put forward a method for detecting early damage to nonlinear systems by combining the NOFRFs with the convolutional neural network and long short-term memory network (CNN-LSTM) model, which was verified through a cantilever beam with a breathing crack. Further studies on the feature extraction method of the WNOFRFs and the associated index K indicated that this method can improve the performance of crack detection. A fault diagnosis method known as the improved WNOFRFs is proposed, which is employed to diagnose rotor cracks. The effectiveness of the method was verified through a mathematical model of the cracked Jeffcott rotor system and an experimental rotor system with crack faults
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
