基于多目标的九轴五联动磨床整机分析与结构优化

Whole Machine Analysis and Structure Optimization of Nine-axis Five-linkage Grinder Based on Multi-objective

现有的九轴五联动数控缓进给磨床运动环节多,在工作过程中可能出现刚度不足、床身等铸件变形及加工精度失准等问题.以磨床的最大变形量以及模态分析下的一阶固有频率作为主要优化目标,整身质量作为次要目标,利用ANSYS Workbench对磨床进行分析与优化,结合灵敏度分析对原有结构进行改进,得到各关键零部件新的设计方案,在磨床整机质量上升一定的程度上,磨床的最大静变形及一阶固有频率得到很大改进.再使用遗传算法(Genetic Algorithm)优化后的极限学习机网络模型(Extreme Learning Machine)联合遗传算法对该九轴五联动磨床改进结构的一些主要参数进行优化:首先,利用灰色关联分析将磨床的整机质量、最大静变形及一阶固有频率转成综合目标灰色关联度,其次,通过遗传算法优化后的极限学习机网络模型(Genetic Algorithm-Extreme Learning Machine)拟合改进磨床主要参数与综合目标灰色关联度之间的非线性耦合关系,最后,使用GA强大的寻优能力在训练好的GA-ELM网络模型中寻找最优工艺参数,优化后九轴五联动磨床的整机质量、最大静变形减小及一阶固有频率相对于改进方案得到优化.为后续相关工艺人员对该磨床的结构和参数优化提供一定理论支撑与参考价值.

The existing nine-axis five-linkage CNC slow-feed grinder has many motion links, which may cause many problems during the working process such as insufficient rigidity, deformation of castings like lathe bed and inaccurate machining accuracy. In this work, the maximum deformation of the grinder and the first-order natural frequency under the modal analysis are taken as the main optimization goal and the overall quality as the secondary goal. ANSYS Workbench is used to analyze and optimize the grinder, and the original structure is then improved by the sensitivity analysis, finally obtaining the new design scheme for each key part. When the quality of the whole grinder is improved to a certain extent, the maximum static deformation and first-order natural frequency of the grinder is greatly improved. Besides, the extreme learning machine network model(Extreme Learning Machine) optimized by genetic algorithm is combined with genetic algorithm to optimize some main parameters of the improved structure of the nine-axis five-linkage grinder. Firstly, the whole machine mass, maximum static deformation and first-order natural frequency of grinding machine are transformed into comprehensive target grey correlation degree via grey correlation analysis. Then, the network model of the extreme learning machine(Genetic Algorithm-Extreme Learning Machine) optimized by the genetic algorithm is used to fit the nonlinear coupling relationship between the main parameters of the grinder and the gray correlation degree of the comprehensive target. Finally, GA′s powerful optimization ability is used to find the optimal process parameters in the trained GA-ELM network model. After optimization, the overall quality of the nine-axis five-linkage grinding machine, the reduction of the maximum static deformation and the first-order natural frequency are optimized when compared with the improved scheme. This method provides a certain theoretical support and reference value for the subsequent technicians to optimize the structure and parameters of the grinder.