Localized defects in ball bearing components would cause additional vibration and it is imperative to reveal the vibration mechanism. The relationship between fault characteristic frequency (fBPFI and fBPFO) and multiple defect parameters of ball bearing were given in this paper. Considering elastohydrodynamic lubrication (EHL), radial clearance, time-varying displacement and excitation force generated from multiple defects, a 4 degree-of-freedom (DOF) dynamic model for ball bearing with multiple defects on inner or outer raceway was established, and the model has been verified by experiments. Vibration signals of ball bearing with different defects parameters were simulated, the effects of the angle between two defects ( θIAD and θOAD), the number of defects ( NDI and NDO) and the location of defects on outer raceway on dynamic response were studied. Comparing simulated signals with experimental results, it is shown that more impulses of acceleration signals are generated by multiple defects than that by single defect, meanwhile time delay due to two defects on raceways could also be found, fault characteristic frequency and their harmonics frequencies appeared in the envelope spectrums. Harmonics frequencies of fBPFI are modulated mainly by 2 fs instead of fs in frequency domain for multiple defects on inner raceway. The amplitudes of fBPFO and fBPFI change as Fourier curve when θOAD and θIAD varied within a certain range, and a series of Fourier function are given to describe the mathematic relationship.

The rotor-bearing-housing system is widely used in rotating machines, which influences the performance of the whole machine. Considering the distribution of rotor mass, the rotor-bearing-housing system with local defect in the outer ring is modeled based on the energy method. In order to make the model agree well with the experimental results, a new rotor mass distribution method was introduced in the modeling process. The simulated vibration signal was obtained by solving the dynamic equations with the Runge-Kutta method. The vibration responses of rotor-bearing-housing system under different distributed disks at both drive end and fan end are simulated. In the simulation results, the outer ring fault signal at the drive end and the vibration signal at the fan end are compared with the experimental signals to discuss the influence of rotor mass distribution on the vibration response of the bearing at both ends of the system. The results show that the simulation signal generated by the dynamic model of the rotor-bearing-housing system with a more uniform rotor mass distribution is more consistent with the experimental signal. The vibration response of the faulty bearing at drive end is compared at different defect sizes, and the variation trend of the amplitude of their characteristic frequency is obtained. This method is helpful for structural optimization and fault diagnosis of the rotor-bearing-housing system.