Research On The Integrated Fast Algorithm Of Bearing Quasi-dynamics And Tehl By Using Neural Networks

As a key component of rotating machinery,the reliability of bearings directly affects the overall performance of the mechanical system.Especially in advanced fields such as aerospace,the aeroengine main shaft bearing will be in more severe working conditions such as high speed,high temperature and heavy load in order to make dn value reach 3.0×106 mm?r/min.Therefore,the material design and reliability analysis methods of the main shaft bearing need to be further innovated and developed on the basis of existing theories.In material genetic engineering,High-throughput experiments and high-throughput calculations can realize the rapid preparation of materials and characterization of physical parameters.As for the reliability analysis of the aeroengine main shaft bearing,the deterioration of lubrication state is the main reason for bearing failure.Combined with Physical properties of lubricating materials,working conditions and other factors,it is important to predict the lubrication state and determine the service performance of the bearing.In order to accurately judge the lubrication state of the bearing,the dynamic analysis and the lubrication analysis of the bearing are indispensable.Although both analysis methods have been gradually improved,it is difficult to integrate two kinds of complex numerical calculation process directly.In addition,for different working conditions,the solution methods of lubrication analysis models are different.Therefore,it is necessary to establish a unified integration framework for bearing dynamic analysis and lubrication analysis.In this paper,the artificial neural network model is established to learn the internal law from the thermoelastohydrodynamic analysis data.Some dimensionless parameters related to speed,load,temperature and material are used as the input of neural network model to predict the minimum film thickness,friction coefficient and other key parameters of lubrication state.The improved particle swarm optimization algorithm is used to optimize the neural network.The neural network model can directly integrate with the bearing quasi-dynamics calculation process instead of the thermal elastohydrodynamic analysis model.At the same time,the database related to the integration method is constructed,and a complete integration algorithm framework is formed.The accuracy of the integration algorithm is verified by an example,which can meet the engineering requirements.In addition,the integration method,directly learning knowledge from the data,avoids the problem of parameter adjustment and convergence in lubrication analysis,and improves the calculation efficiency of the combination of quasi-dynamics and thermoelastohydrodynamic lubrication analysis.