| H13 hot work die steel has high hardness,good wear resistance,hardenability and impact toughness,which is widely used for manufacturing hot forging die,die casting die and hot extrusion die Die failure is closely related to the distribution of the residual stress on the machined surface.The residual tensile stress will intensify the surface crack initiation and lead to die failure during die service,while the compressive residual stress has the opposite effect.Predictive research of the residual stresses on machined surface of mold and regulate residual stress distribution has great significance and practical value to improve the quality of machined surfaces and extend the service fatigue life of mold.In this paper,a series of researches are conducted around the hybrid modeling of residual stresses prediction on the machined surface of H13 steel,the main contents are as follows:Firstly,the finite element model of H13 steel orthogonal cutting is established by ABAQUS,to obtain the physical field datas of the machined surface such as stress,strain and temperature.The stress relaxation analytical model,which consider the elastic-plastic state of the material,is used to calculate residual stresses of the machined surface.The data transfer between the basic physical fields of the machined surface and the stress relaxation analytical model is realized by the secondary development of ABAQUS based on Python.And then the combined model of residual stresses on the machined surface is established.The credibility of the combined model is verified by comparing the experimental values of residual stresses.Then,based on the combined model of residual stress,the effect of tool structure parameters(tool rake angle,clearance angle,edge radius)and cutting parameters(cutting speed,depth of cut)on the residual stress on the machined surface of H13 steel is studied.The evolution of cutting forces and the temperature distribution of the machined surface are analyzed,which under the effect of different tool structure parameters and cutting parameters,and the distribution of residual stresses on the machined surface of H13 steel is studied.Through orthogonal experimental design,the primary and secondary impact of tool structure parameters and cutting parameters on the characteristic values of residual stress distribution curve(maximum residual tensile stress and maximum residual compressive stress)on the machined surface of H13 steel is studied.Finally,a hybrid modeling prediction model of residual stresses on machined surface of H13 steel based on Sparrow Search Algorithm(SSA)with improved BP neural network(SSA-BP)is proposed,which is combining finite element numerical simulation,stress relaxation analytical model and machine learning.The input parameters are the tool rake angle,clearance angle,edge radius.cutting speed and depth of cut,and the output parameters are the maximum residual tensile stress and maximum residual compressive stress.The SSABP hybrid prediction model is trained by acquiring data sets from combined model of residual stress.By comparing with the traditional BP neural network and genetic algorithm optimized BP neural network(GA-BP),and through the analysis of error index,the SSA-BP model is found to be flexible and efficient in predicting the characteristic values of the residual stress distribution curve on the machined surface of H13 steelBy establishing the hybrid prediction model for the machined surface residual stresses and conducting prediction research and analysis,it not only helps to explain the influence of tool structure parameters and cutting parameters on the residual stress distribution,but also provides a reference for the development of efficient,low-cost and high-precision residual stress prediction methods for the machined surface. |
