基于随机森林的钛加工表面质量评定研究  

Study of Surface Quality Evaluation of Titanium Processing Based on Random Forest

作　　者：李晟 周超[1] LI Sheng;ZHOU Chao(School of Mechanical Engineering and Automation,Fuzhou University,Fuzhou 350116,China)

机构地区：[1]福州大学机械工程及自动化学院,福建福州350116

出　　处：《机械制造与自动化》2020年第6期36-38,60,共4页Machine Building & Automation

基　　金：福建省自然科学基金项目(2016J01225)。

摘　　要：针对不同工艺加工的零件表面质量未有统一评价标准且三维参数众多未有统一规范的问题,提出了一种基于随机森林算法筛选用以评定表面质量的三维参数方法。采用三维轮廓仪,分别测量采用超声滚压、车削、磨削、抛光、铣削5种工艺加工的钛材表面形貌,基于随机森林算法,通过学习选用的高度类及间距类共9个三维参数作为研究特征,以达到准确分类表面的目的,并筛选出用于评定钛材加工表面质量的参数集。研究表明:基于随机森林算法,算术平均高度、均方根高度及偏斜度在分类过程中有最高的重要性,且在三者组成的三维空间中表面区分效果良好,作为参数集能够合理、可靠地评定钛材加工表面质量。Aiming at the problem that there is no uniform evaluation standard and specification for surface quality of mechanical parts machined by different processes and 3D parameters,a method based on random forest algorithm used to screen 3D parameters for surface quality evaluation is proposed in this paper.The surface topography of titanium material processed by ultrasonic rolling,turning,grinding,polishing and milling is measured by 3D surface profilometer.Through learning,nine 3D parameters for height class and distance class are selected as the research features,which are used to classify different machined surfaces accurately and screen out 3D parameter sets for evaluating the surface quality of the titanium material.The result shows that based on the random forest algorithm,the arithmetic average height,root mean square height and skew are of the highest importance in the classification process,and the effect for the surface discrimination is good in the 3D space composed of them,which can be used to reasonably and reliably evaluate the processing surface quality.

关 键 词：表面形貌 随机森林 特征重要度 参数评定 

分 类 号：TH161.1[机械工程—机械制造及自动化] TP301.6[自动化与计算机技术—计算机系统结构]