应用差分进化算法评定椭圆轮廓度误差  

Differential Evolution Algorithm for Evaluation of the Elliptical Profile Error

作　　者：易建[1,2] 

机构地区：[1]重庆工商大学制造装备机构设计与控制重庆市重点实验室,重庆400067 [2]泸州职业技术学院机械工程系,四川泸州646005

出　　处：《机械设计与制造》2016年第9期61-63,68,共4页Machinery Design & Manufacture

基　　金：重庆市前沿与应用基础研究计划项目(cstc2015jcyjA70006);重庆市教育委员会科学技术研究项目(KJ1403201);四川省教育厅重点科技计划项目资助(14ZA0333)

摘　　要：根据评定误差的最小包容区域准则(Minimum zone criteria,MZC),应用坐标变换法建立评定椭圆轮廓度误差的5变量鞍点规划模型。因MZC误差评定模型的关键是计算每个测点到理想椭圆轮廓的最小距离,为此采用一维搜索算法求解该最小距离。由于差分进化(Differential evolution,DE)算法具有概念简单和收敛速度快的优点,文中利用该算法求解评定椭圆轮廓度误差优化问题。给出2个椭圆轮廓度误差评定实例。结果表明,提出的模型和算法可行有效,其评定结果小于应用最小外接椭圆和最大内接椭圆法求得的误差。在相同计算开销的条件下,DE算法的性能指标优于遗传算法和粒子群算法。A saddle point programming model with 5 variables is constructed to evaluate elliptical profile errors by using the coordinate transformation method according to the minimum zone criteria （ MZC ） for evaluating errors. The key of the model for evaluating errors based on MZC is to calculate the minimum distance from every measuring point to the ideal elliptical profile, hence a one-dimensional search algorithm is used to solve this distance. Difference evolution （DE） algorithm is characterized by simple concept and quick convergence, thus this algorithm is employed to solve the optimization problem for evaluating elliptical profile errors. Two study cases for evaluation of elliptical profile errors are given to show that the proposed model and algorithm is feasible and effective and its evaluating results are less than ones obtained by the minimum circumscribed ellipse and maximum inscribed ellipse method. DE algorithm is superior to genetic algorithm and particle swarm optimization in term of perfeormance indexes, under the condition of same computational cost.

关 键 词：椭圆轮廓度误差 误差评定 鞍点规划 一维搜索 差分进化算法 

分 类 号：TH16[机械工程—机械制造及自动化] TH161.13