Study On Accelerated Life Test Of Flexible Thin-walled Bearing Of Harmonic Reducer

Flexible thin-walled bearings are the core components of harmonic reducers,and their performance determines the performance of harmonic reducers.The flexible thin-walled bearing consists of an inner ring,an outer ring,rolling elements and a cage.When the flexible thin-walled bearing is assembled to the harmonic reducer,the inner ring is mounted on the elliptical shaft and occurs “permanent deformation”.The flexible thin-walled bearing is deformed into an elliptical shape as a whole.During the operation of the harmonic reducer,the outer ring of the flexible thin-walled bearing undergoes cyclic elastic deformation,which not only bears the cyclic stress load of the rolling element,but also withstands the cyclic alternating extrusion from the long axis of the ellipse.Compared to ordinary rolling bearings,flexible thin-walled bearings are more prone to fatigue failure.In harmonic deceleration,the failure of flexible thin-walled bearings will cause the harmonic reducer to fail to operate normally.In this paper,the life detection of flexible thin-walled bearings is systematically studied from four aspects: construction of flexible thin-wall life test bench,design of accelerated life test scheme,fault signal extraction and parameter estimation method of Weibull distribution.The main research contents are as follows:(1)By analyzing the life testing requirements of flexible thin-walled bearings and simulating the load distribution of flexible thin-walled bearings under harmonic retarder conditions,a flexible thin-walled bearing life test bench with a flexible thin-walled bearing as a single test object was designed.And the loading mechanism,drive shaft assembly,assembly mechanism,electrical control system and signal acquisition system of the test bench were introduced in detail,which provided conditions for the accelerated life test of flexible thin-walled bearings.(2)With reference to the accelerated life test of ordinary rolling bearings,an accelerated life test program for flexible thin-walled bearings is designed for the characteristics of flexible thin-walled bearings.The preconditions for the accelerated life test of the flexible thin-walled bearing,the selection of the accelerated load,the test method,the test conditions,and the operation flow are described in detail.Since the correction parameters for the flexible thin-wall bearing life formula at home and abroad are mainly concentrated in the theoretical research,it is proposed that the correction coefficient of the flexible thin-wall bearing life formula is obtained by analyzing the life test data,and the completed theoretical basis and calculation method are given.(3)In the life test of flexible thin-walled bearings,how to judge whether the bearing is faulty is the key to the accelerated life test method of flexible thin-walled bearings.The time domain signal characteristic parameter calculation software is used to calculate the vibration signal collected by the accelerated life test,and the time domain characteristic parameter values such as the mean value,root mean square value,variance and kurtosis of the vibration signal at each stage of the life test are obtained.By comparing the changes of the time domain characteristic parameters of the vibration signals at different stages,it is found that the value of the root mean square will decrease rapidly after the failure of the flexible thin wall.Therefore,the value of the root mean square is chosen as the stop reference indicator.The vibration signal of each stage of the accelerated life test is processed by the wavelet analysis algorithm,and the fault signal is extracted as a stop reference.(4)By analyzing the difference between the two-parameter and three-parameter Weibull distribution,the limitations of the two-parameter Weibull distribution are pointed out.The parameter estimation of three-parameter Weibull distribution is performed by using maximum likelihood estimation,ant colony algorithm and particle swarm optimization.Taking the samples with the number of samples 100 and 20 as examples,the correlation coefficient,the AD and OSL values of the AD test method,and the P-P map are drawn as evaluation indexes to evaluate the parameter estimation effects of each algorithm.