Based on the vibration time series during the life cycle of rolling bearings

a maximum entropy Poisson evaluation model was constructed to study the evolution process of rolling bearing vibration performance. The vibration time series were divided into different segments

and then the variation probability

vibration performance maintaining reliability and their variation speed and acceleration were calculated based on maximum entropy principle and Poisson process for each vibration time series in relation to the intrinsic series. The relationships between performance variation indexes and sample size were analyzed to select the appropriate sample size. Dynamic average uncertainty was used to analyze the uncertainty of evaluation results of performance maintaining reliability. Result showed that the sample sizes of 800-1000 and 500-900 were selected respectively

for case 1 and case 2

which can make the intrinsic series data sample contain enough vibration information

but also effectively evaluate the specific variation process of bearing vibration performance.