| Five-axle CNC machine tool is the most commonly used method to solve the problem of complex surface parts,and it also provides an unparalleled high-efficiency machining method for aero-engine blades,heavy-duty generator rotors and other parts.Five-axle CNC creep-feed grinder can not only grind parts with high precision and minimal surface roughness,but also with high efficiency and strong grinding.It is widely used for machining parts with high precision and complex shape.However,to achieve efficient and precise one-time forming of complex parts,the grinder needs good dynamic and static characteristics,which puts forward high level requirements for the design of the grinder.In this paper,the existing 9-axle 5-link CNC creep-feed grinder has many moving links,and its rigidity requirement is difficult to guarantee the deformation of casting such as lathe body,inaccurate machining accuracy,multiple design objectives in the process of grinder design,and the improvement and optimization need to be verified repeatedly,which makes the production research and development cycle long.A method for dynamic and static characteristics analysis and structure optimization of 9-axis 5-link grinder was studied.Through finite element analysis software,the dynamic and static characteristics of key parts of grinder and the whole machine are analyzed,improved and optimized,and the prediction model of dynamic and static characteristics of the machine tool is established based on neural network based on Adam algorithm,which makes the original complex and time-consuming analysis and verification link well simplified and reduces the design and development cycle.The main contents of this paper are as follows:(1)The parametric model of the grinding machine is established.The three-dimensional model is simplified reasonably to reduce the time cost of finite element analysis,and the structural dimension parameters are selected as the variables for subsequent analysis.Then the Ansys Workbench software is used to analyze the dynamic and static characteristics of the key parts of the machine tool and the whole machine,and the sensitivity analysis of each parameter is carried out,which verifies the effectiveness of each selected parameter.(2)Combined with the finite element analysis and sensitivity analysis,the structure of the original structure is improved,and the new design scheme of the key parts is obtained.The structural parameters of the improved part are selected as the variables of the subsequent multiobjective optimization of the whole machine,and then the dynamic and static characteristics of the whole machine after the parts are improved are analyzed,The results show that the performance of the machine tool will be improved to a certain extent when only the parts are optimized.(3)Taking the first-order natural frequency and maximum static deformation of the whole machine as the optimization objective function and its quality as the constraint condition,the multi-objective optimization theory and response surface method are used to optimize the design parameters of the improved part of the whole machine.By comparing the three sets of solutions obtained from the optimization,the final optimization scheme is selected,and the effectiveness of the multi-objective optimization is verified.(4)Combined with the sample set established in the optimization process,the neural network algorithm is used to establish the dynamic and static characteristic prediction model of machine tool,which is used to predict the dynamic and static characteristic results of machine tool improvement scheme under different parameter selection.Through random comparison and verification,the prediction error of the prediction model is small,and the time cost is low,which can provide a good reference for the actual machine tool optimization. |
PDF Full Text Request