Fault size estimation is of great importance to bearing performance degradation assessment and life prediction. Until now, fault size estimation has generally been based on acoustic emission signals or vibration signals; an approach based on current signals has not yet been mentioned. In the present research, an approximate estimation approach based on current is introduced. The proposed approach is easy to implement for existing inverter-driven induction motors without complicated calculations and additional sensors, immune to external disturbances, and suitable for harsh conditions. Firstly, a feature transmission route from spall, to Hertzian forces, and then to friction torque is simulated based on a spall model and dynamic model of the bearing. Based on simulated results, the relation between spall size and the multiple characteristic vibration frequencies in friction torque is revealed. Secondly, the multiple characteristic vibration frequencies modulated in the current is investigated. Analysis results show that those frequencies modulated in the current are independent of each other, without spectrum overlap. Thirdly, to address the issue of which fault features modulated in the current are very weak, a fault-feature-highlighting approach based on reduced voltage frequency ratio is proposed. Finally, experimental tests were conducted. The obtained results validate that the proposed approach is feasible and effective for spall size estimation.

No abstract

No abstract

A new method for identifying induction motor bearing fault is introduced in this paper, it's based on the Volterra series which can describe the nonlinear transfer characteristics of system. Firstly, analyze the theory that bearing fault can cause torque vibration, and the simplify equation of stator current and voltage on bearing fault state is derived. The stator voltage and current signals are used as the input and output of Volterra series, then adaptive chaotic quantum particle swarm optimization (ACQPSO) is introduced for the identification of Volterra series time-domain kernel, and the bearing fault can be identified by the changes of nonlinear transfer characteristics. In order to validate the method, the induction motor bearing fault simulated test system is established in the lab to simulate the single point damage of bearing outer race which gradually expand; through the extraction of the changes of the kernel, the bearing fault and its severity can be identified. Thus verified the feasibility and effectiveness of the proposed method, the method is suitable for the prediction of the trends of bearing fault.

No abstract