Prediction Of Elastic Deformation In Thin-walled Blade Milling And Research On The Error Control

The thin-walled blade of complex curved surface is the core part of aeroengine.Its machining accuracy and surface quality have an important influence on the performance of the engine.At present,five axis NC milling is a widely used form of machining.However,because the blade has the characteristics of complex surface and weak rigidity,the milling force will cause elastic deformation during the milling process,which leads to the deviation between the theoretical surface and the actual surface after machining.Therefore,it is of great significance to study the prediction of deformation errors and corresponding control methods in the process of blade milling.In this paper,the machining process of thin-walled blade milling is taken as the research object.The finite element numerical calculation and BP neural network are used to study the deformation prediction and error control of blade machining.The research work is mainly reflected in the following aspects:(1)Based on DEFORM-3D software,a three-dimensional thermo mechanical coupled finite element model for ball end milling process is established.The orthogonal test of milling is designed,taking milling parameters as variables and milling force as the test index.The effect of milling parameters on milling force is studied by using range analysis and dominance analysis.The results provide reference for choosing cutting parameters in the process of blade deformation calculation.(2)Based on the basic principles of material mechanics and elasticity,the combined bending and torsion analysis of the simplified blade model is carried out.In order to improve the calculation accuracy of deformation data,a coupling iteration scheme between milling force and elastic deformation is established.The deformation of discrete blade contact points on blade surface is calculated by ANSYS finite element software and iterative scheme.Based on the surface interpolation algorithm,the blade surface deformation error model is obtained through fitting the deformation data by MATLAB,and the deformation law of the blade is analyzed.(3)Based on BP neural network,the prediction model of blade deformation error was established.The function of the prediction model is to input the three-dimensional coordinates at a certain cutter location and output the corresponding deformation data.According to the function and precision requirements of the BP network prediction model,the BP neural network with the best performance is selected as the error prediction model for the blade process ing by changing the relevant parameters to train many times.Using MATLAB GUIDE integrated development environment,a graphical user interface for prediction model construction and blade deformation prediction is designed.(4)Based on the principle of mirror error compensation,the corresponding compensation scheme is formulated for the error of the elastic deformation produced during the milling process of the blade,and the GUI interface of the error compensation is designed.For the problem of large change in the blade axis vector of the blade edge,the tool axis vector optimization scheme is formulated by combining the angular velocity constraint conditions of the machine tool axis and the vector smoothness of the cutter shaft,and the effectiveness of the scheme is analyzed by an example.Finally,the BP network prediction model construction,the deformation error prediction and the error compensation GUI interface are integrated,and the visual operation of the elastic deformation prediction and error compensation of the thin-walled blade milling is realized.