|
The surface roughness is one important indicator to measure the machining surface quality predicted by modeling roughness, to provide the basis for the optimization of milling parameters to accurately predict the surface roughness of the workpiece , before the actual processing , in order to improve the surface quality of the product , production cost savings . Predict the geometry model, BP network model , regression analysis model through the establishment of the milled surface roughness , and the milling surface roughness of titanium alloy TC4 carried milling pilot study , surface roughness , and test data derived by processing parameters given forecast comparative analysis . The thesis main research work as follows: 1, considering the direction of the inclination of the milling cutter , the inclination angle , the feed mode, and the influence of the eccentricity of the rotary spindle , the axial motion , and other factors on the processing surface , ball head cutter , for example , the use of coordinates transformation principle and matrix algorithms to derive a ball end mill multi-axis milling process at any point on the cutting edge in the equation of the locus of the workpiece coordinate system , in order to establish a geometric model of surface roughness . To cutting speed , feed rate , radial depth of cut three factors as variables , empirical model of surface roughness prediction using neural networks and regression analysis . Proven neural network forecasting model , and regression analysis model prediction accuracy , generalization ability , can be used to predict the milling parameters on the machined surface roughness effects reveal the variation of the machined surface quality with the milling parameters , cutting parameters preferred and surface quality control to provide the basis . Titanium alloy TC4 three factors and three levels orthogonal milling test , and the test results were very poor , provide training samples for the prediction model , at the same time , the comparative analysis of the feasibility and accuracy of the research model . Roughness model established in this paper , after the validation of the test data , accurate prediction , the prediction results can meet the requirements of the actual processing errors .
|