基于SVR-KM算法的一种立式加工中心Y轴运动直线度一致性研究  

Research on the consistency of Y-axis motion straightness of vertical machining center based on SVR-KM algorithm

作　　者：申鹏程 贾书海[1] 杨红军 常艳 陈花玲[1] 梅雪松[1] Shen Pengcheng;Jia Shuhai;Yang Hongjun;Chang Yan;Chen Hualing;Mei Xuesong(School of Mechanical Engineering,Xi’an Jiaotong University,Xi’an 710049,China;Bochi Machine Tool Group Co.,Ltd.,Baoji 721000,Shaanxi,China)

机构地区：[1]西安交通大学机械工程学院,西安710049 [2]宝鸡忠诚机床股份有限公司,宝鸡721000

出　　处：《现代制造工程》2019年第11期79-84,78,共7页Modern Manufacturing Engineering

基　　金：国家科技重大专项资金资助项目(2015ZX04005001)

摘　　要：针对某立式加工中心的制造一致性问题,采用非参数统计中的Kruskal-Wallis检验,分析了影响Y轴运动直线度的因素,提出了基于支持向量回归机(Support Vector Regression,SVR)的Y轴运动直线度精度区间预测算法,通过遗传算法对支持向量回归机的惩罚函数参数C和高斯核函数参数gamma进行了优化,使算法具有了更高的预测精度和更好的适应性;在对精度区间准确预测的基础上,通过KM(Kuhn-Munkras)算法对机床底座和装配人员进行二分匹配,显著提高了Y轴运动直线度的一致性。结果表明:采用支持向量回归机预测算法在置信度为90%的情况下其预测的精度区间宽度为3μm,蒙特卡洛模拟显示一致性提升了47%,可为提高立式加工中心制造一致性提供新思路。For the manufacturing consistency problem of a vertical machining center,the Kruskal-Wallis test in non-parametric statistics was used to analyze the factors affecting the Y-axis motion straightness.A prediction algorithm of the Y-axis motion straightness based on the Support Vector Regression(SVR)was proposed.The genetic algorithm optimized C and gamma in Gaussian kernel parameters of SVR,which made the algorithm has higher prediction accuracy and better adaptability.On the basis of accurate prediction of accuracy interval,the KM(Kuhn-Munkras)algorithm was applied to the bipartite graph matching for the base of the machine tool and the assembler,significantly improved the consistency of the Y-axis motion straightness.The results show that the SVR prediction algorithm has an interval width of 3μm at a confidence level of 90%,and Monte Carlo simulation displayed a 47%improvement in consistency,which provided new ideas for improving the consistency of manufacturing of vertical machining centers.

关 键 词：Kruskal-Wallis检验 支持向量回归机 KM算法 精度一致性 

分 类 号：TH186[机械工程—机械制造及自动化] N945.12[自然科学总论—系统科学]