Experimental Assessment Of Rolling Bearing Performance Based On Poor Information Process

The dynamic performance which contains vibration and friction torque isparticularly important in the rolling bearing performance indicators. The bearingvibration can accelerate the wear and fatigued damage of rolling bearing components,and friction torque is an important indicator of evaluating the operational sensitivity ofrolling bearing. The investigation shows that vibration and friction torque of rollingbearing belongs to an information poor system with unknown probability distributionsand trends and nonlinear characteristics. This counteracts dynamical assessment forthe characteristics of the rolling bearing vibration and friction torque as a time series.Therefore, how to analyse effectively the complex vibration and friction torquecharacteristics has become a problem to be solved.Firstly, the chaos theory is employed to recover the original dynamic characteristicsof a vibration or friction torque time series by means of the phase space reconstructiontheory. Secondly, the probability density function of the current sample and the priorsample are established by the maximum-entropy bootstrap method in the phase space.Thirdly, the dynamical Bayesian probability density function of the vibration orfriction torque characteristic parameters is constructed by the Bayesian statisticaltheory. Finally, the method for point estimation, interval estimation, and trendestimation of the characteristic parameters is proposed. At this point, the informationpoor mathematical model is constructed completely. Then the problems of evaluatingthe rolling bearing vibration and friction torque can be solved effectively by theinformation poor mathematical model.By evaluating of the rolling bearing vibration experimental data with four differentdamaged point diameter, the result shows that the sample mean X of the vibration isperfectly enveloped by the estimated interval [XL, XU], and the error between theestimated true value X0and the sample mean X is very small, and the vibration meanuncertainty Umeangoes up with the increase of the rolling bearing damaged point diameter.By evaluating of the rolling bearing friction torque experimental data with threedifferent rotational speed, the result shows that the sample mean X of the frictiontorque is perfectly enveloped by the estimated interval [XL, XU], and the error betweenthe estimated true value X0and the sample mean X is very small, and the frictiontorque mean uncertainty Umeangoes up with the increase of the rolling bearingrotational speed.Elementary theories and methods studied in this paper are of the versatility andfeasibility. The investigation shows that the proposed mathematical model isconfirmed to be reasonable and feasible with the evaluation of rolling bearingvibration and friction torque experimental data. The error between the calculated resultand the experimental result is very small and requirements in engineering are satisfied.Thus it lays a new theoretical foundation for the evaluation of nonlinear dynamicscharacteristic evolution of the rolling bearing performance.