Design And Optimization Of Vortex Suppression Structure For Hydrostatic Bearing

Hydrostatic bearings are important components commonly used in precision machine tools,and their performance directly affects the operating efficiency and life of machine tools and equipment.During the operation of the bearings,vortex phenomena are generated due to fluid dynamics effects,resulting in increased bearing noise,vibration and losses.In this paper,the vortex suppression problem of hydrostatic bearings was investigated,a double-slit structure was designed,and the bearing performance and flow patterns were analysed using unstructured meshing methods and CFD simulation techniques.Further,a deep neural network model was constructed and the prediction of the vortex state of the hydrostatic bearing fluid in the double-slit structure was achieved by training a large amount of simulation data.The main contents and contributions are as follows:(1)An unstructured meshing method was proposed,which makes the mesh skewness of the oil film region and other regions less than 0.4 and 0.8 respectively,ensuring the accuracy and efficiency of the simulation.Further,CFD simulation was used to obtain parameters such as support load capacity,stiffness and damping to analyse the performance of the hydrostatic bearing.The average error between the simulation results and the actual value is 10.76%,which was better than the results calculated by the traditional empirical formula.(2)The flow pattern inside the bearing was visualised and analysed,and the double-slit structure was found to be effective in suppressing the formation of vortex at an oil film movement speed of 400mm/s.In addition,this study obtained the optimum double-slit structure parameters,i.e.double-slit structure width 0.1mm,double-slit structure height 0.1mm and sub oil cavity height 0.1mm,by observing the vortex state through single-factor tests.(3)Based on the results of single-factor simulation experiments,a vortex quantity dataset was produced,a deep neural network model was constructed with the double-slit structure parameters and relevant simulation parameters as input and the vortex quantity as output,and the neural network was trained and tested using this dataset and compared with other regression algorithms.The results show that the neural network has high prediction accuracy and generalisation capability,with a performance index R~2of 0.9022 and MSE of 1.2242.