基于光散射法和GA-SVM的高速钢轧辊粗糙度检测  

Analysis and Optimization of Measuring Roughness Grade of High Speed Steel Roller by Light Scattering Method

作　　者：李健 吴怀超[1] 赵丽梅[1] 杨绿[1] LI Jian;WU Huai-chao;ZHAO Li-mei;YANG Lv(School of Mechanical Engineering,Guizhou University,Guiyang 550025,China)

机构地区：[1]贵州大学机械工程学院,贵阳550025

出　　处：《组合机床与自动化加工技术》2022年第10期97-99,104,共4页Modular Machine Tool & Automatic Manufacturing Technique

基　　金：贵州省科技重大专项(黔科合重大专项字[2019]3016号);贵州省科技创新人才团队项目(黔科合平台人才[2020]5020);贵州大学培育项目(贵大培育[2019]02号);贵州省高层次创新型人才培养计划项目(黔科合平台人才[2016]5659)。

摘　　要：针对传统接触式检测高速钢轧辊粗糙度操作复杂、效率低的问题,提出一种利用光散射法非接触检测粗糙度的技术方案,通过搭建平台采集不同粗糙度高速钢轧辊的光散射图像,利用MATLAB提取图像特征值,将提取出的特征值作为遗传算法优化支持向量机分类的训练集与测试集并查看分类准确率,通过对比四种优化算法支持向量机,结果发现利用遗传算法优化支持向量机粗糙度分类运行时间短、准确率高,可以用来非接触式检测高速钢轧辊的粗糙度。In order to solve the problems of complex operation and low efficiency of traditional contact detection of roughness of high-speed steel rolls,a technical scheme of non-contact roughness detection by light scattering method is proposed.Light scattering images of high-speed steel rolls with different roughness are collected by establishing a platform,and image characteristic values are extracted by MATLAB.The extracted eigenvalues are used as training set and test set of genetic algorithm to optimize the classification of support vector machine and the classification accuracy is checked.By comparing the four optimization algorithm support vector machines,it is found that the genetic algorithm can be used to optimize the roughness classification of support vector machines with short running time and high accuracy,which can be used to non-contact detect the roughness of high-speed steel rolls.

关 键 词：光散射法 表面粗糙度 遗传算法 支持向量机 

分 类 号：TH16[机械工程—机械制造及自动化] TG502[金属学及工艺—金属切削加工及机床]