Thin-walled parts are widely used in military aviation and other fields.Due to the special shape of thin-walled curved parts,the machining accuracy and deformation control requirements are very strict.The redistribution of residual stress after machining is an important reason affecting the quality of finished parts.In this paper,the 7075-T651 aluminum alloy annular thin-walled parts were taken as the analysis object,and the surface milling model were established by ABAQUS simulation software.The effects of milling speed,radial depth,axial depth and feed per tooth on the residual stress distribution of the annular thin-walled parts were analyzed by single factor anal-ysis.The results show that the maximum residual stress can be reduced by 13％～15％ with high speed and small cutting depth.Based on the simulation study,the relationship between the maximum residual stress and the process parameters was established by BP neural network and GRNN neural network,so as to realize the prediction of the maximum residual stress and improve the milling quality.The results show that the minimum error of BP network training results is 2.3％,and the minimum error of GRNN neural network prediction is 0.07％.After comprehen-sive comparison,GRNN neural network has better prediction ability.
维普
万方数据