基于组合代理模型的零件表面形貌构建
Surface Topography Construction of Parts Based on Combinatorial
郑林青 1唐文斌 1陈永当 1吕英豪 1卢媛媛1
作者信息
- 1. 西安工程大学机电工程学院,西安 710048
- 折叠
摘要
针对当前零件表面形貌建模存在模型构建准确度低、需要测量点数量多的问题,提出一种基于组合代理模型的零件表面形貌构建方法.在模型构建方面,分别提出基于克里金模型和径向基函数模型的表面形貌构建方法,给出组合两种代理模型的整体思路,并引入交叉验证策略建立组合代理模型权重因子的确定方法.在实验方面,采用组合代理模型分别针对仿真数据和实际测量数据进行拟合,并以均方根误差、均方根标准预测误差为指标对组合代理模型进行评价.实验验证结果表明所提方法相对单一模型在零件表面形貌建模问题上模型精度更高,所需要测量点数更少.
Abstract
The surface topography of a part is an important feature reflecting the machining quality of the part,and its accurate construction is the basis for evaluating the machining error and optimizing the machi-ning process.Aiming at the problems of low accuracy of model construction and high number of measure-ment points in current part surface topography modelling,this paper proposes a method of part surface to-pography construction based on combined agent model.In terms of model construction,the surface topogra-phy construction method based on Kriging model and radial basis function model is proposed respectively,the overall idea of combining two kinds of agent models is given,and the cross-validation strategy is intro-duced to establish the method of determining the weight factor of the combined agent model.In the experi-mental aspect,the combined agent model is fitted to the simulation data and the actual measurement data re-spectively,and the root mean square error and the root mean square standard error are used as the indicators to evaluate the combined agent model.The experimental validation results show that the method proposed in this paper has higher model accuracy and requires fewer measurement points than a single model for the part surface topography modelling problem.
关键词
代理模型/表面形貌/交叉验证/预测精度Key words
agent model/surface topography/cross-validation/prediction accuracy引用本文复制引用
基金项目
国家自然科学基金(52105559)
航空科学基金(2022Z050111001)
西安市科技计划(23GXFW0022)
出版年
2024
NETL
NSTL
万方数据