Simulation And Experimental Researchs On Cylindrical Grinding Of 18CrNiMo7-6 Steel

As a precision machining method,grinding has the advantages of good surface quality,high machining accuracy,and good self-sharpening.Therefore,it is often used as the last machining process and widely used in all areas of processing.Compared with cutting methods such as turning and milling,the shape of the grinding wheel,used in grinding is more complicating.It’s difficult to research the grinding mechanism.The research of grinding mechanism is usually carried out by studying the influence of grinding process on grinding force,surface roughness and residual stress of the workpiece.In recent years,computer simulation technology has achieved rapid progress,and there are more options for the research methods of grinding mechanism than experiments.The current grinding simulation technology has a few applications in cylindrical grinding,and often use two-dimensional models or replace them with plane models.Therefore,grinding simulation requires further exploration on the threedimensional scale.In addition,the application of the popular neural network model to the prediction of grinding surface roughness will play a guiding role in the formulation of grinding process.Based on the Abaqus software,this article used Python language for secondary development to simplify the simulation process for the actual needs of cylindrical grinding simulation.The simulation of external grinding force was carried out by using the developed external grinding simulation plug-in program.Cylindrical grinding experiment used 18 Cr Ni Mo7-6 gear steel,which with excellent mechanical properties.The experiment explored the influence of cylindrical grinding process parameters on grinding force,surface roughness and residual stress.The grinding force,which was obtained from the test,and which was compared with the simulation results to verify the feasibility of the simulation method.The GA-BP neural network was established by using the surface roughness datas of the workpiece measured by the experiment.The surface roughness of the workpiece is predicted through the training of the grid.The results are as follows:(1)This article created the Abaqus cylindrical grinding simulation platform program,realized the parametric modeling of multi-abrasive grinding wheels and workpieces.Besides,it integrated the pre-processing functions,such as assembly,meshing,assigning material properties,and creating sets.By running this plug-in in Abaqus,the program greatly simplified the external grinding simulation operation.Based on this program,the external cylindrical grinding simulation was carried out.The simulation results are consistent with the experimental results.With the increase of the process parameters,the grinding force also increases,and the influence of the workpiece speed is the least obvious.Compared with the experiment,the grinding force obtained by simulation has an average error of 18.2%,the error of normal grinding force is less than that of tangential grinding force.(2)In the single factor test of longitudinal cylindrical grinding,the surface roughness of the workpiece is not monotonic with the change of process parameters.When the process parameters increase,it shows a trend of first decreasing and then increasing;with the increase of grinding depth and longitudinal feed speed,the effect of surface residual stress and tensile stress in surface residual stress increases,and the change of workpiece speed has no obvious law on the effect of residual stress.(3)The initial weights and thresholds of BP neural network are optimized by genetic algorithm,and use GA-BP neural network to establish a surface roughness prediction model.The prediction results show that the GA-BP network can meet the accuracy requirements of surface roughness prediction in cylindrical grinding.