Based on grey prediction and bootstrap resampling

a model of dynamic assessment and diagnosis is proposed to evaluate and improve vibration of rolling bearings.The model characterizes vibration of rolling bearings by developing six parameters

viz.dynamic uncertainty

estimated interval

estimated true value

mean uncertainty

mean true value

and systematic error measure.Experiments on HKRB bearing show that the model can separate systematic errors with a minimum of the mean uncertainty

without any requirements for both the probability distribution of random errors and the type of systematic errors.Reliability of assessment is proved to be 100%.The model also discovers the effect of error character on vibration of rolling bearings and then lays a foundation of control over vibration of rolling bearings in the process of production

by revealing error root from making the parts of rolling bearings.