LAUSR

The response of a nonlinear system may present strong resonance under a periodic excitation. Based on the nonlinear system resonance, a method of bearing fault diagnosis is proposed in this paper. The system resonance is generated by finding the proper system parameters after the characteristic signal inputting the Duffing system. Through the resonance, the fault characteristic is amplified and easy to recognize. However, the resonance effect is different in several typical systems. Thus, three intuitional studies on the undamped, underdamped, and overdamped Duffing system are performed to select the optimal nonlinear system. The results show that the undamped Duffing system is superior to the other kind of systems in the weak fault feature extraction. It can not only amplify the fault characteristic but also suppress other frequency components in the spectrum. Meanwhile, the improved quality factor can be used to determine the resonance induced by the external excitation. The accuracy of bearing fault diagnosis is improved greatly. Both numerical simulations and experimental verifications certify the correctness and validity of the method. System resonance method has a certain reference value in enhancing a weak signal. It provides a new idea in weak signal extraction. It can be used in weak characteristic information extraction and other related fields.