机器学习辅助下的五轴数控铣削刀轨优化  

Optimization of Five-axis CNC Milling Tool Paths Assisted by Machine Learning

作　　者：王小刚[1] 邱磊[2] WANG Xiao-gang;QIU Lei(School of Traffic Engineering, Yangzhou Polytechnic Institute, Yangzhou Jiangsu 225127, China;School of Mechanical Engineering, Ningbo Institute of Technology, Ningbo Zhejiang 315336, China)

机构地区：[1]扬州工业职业技术学院交通工程学院,江苏扬州225127 [2]宁波工程学院机械工程学院,浙江宁波315336

出　　处：《组合机床与自动化加工技术》2021年第2期110-113,共4页Modular Machine Tool & Automatic Manufacturing Technique

基　　金：宁波市自然科学基金资助(2016A610106);国家级“中高职衔接专业教师协同研修”项目资助(2018G14)。

摘　　要：针对铣削加工过程中刀具挠度变形的自动补偿问题,提出了一种用于五轴数控加工的刀轨自优化方法。首先,该方法从铣削加工材料去除仿真中获得工艺条件,且将计算出的切削条件与相应的形状误差测量相关联;其次,采用基于统计学习理论的支持向量回归(Support Vector Regression,SVR)来预测所产生的形状误差,并进行自优化和生成刀具路径;最后,在五轴CNC机床上进行了应用,且在两个凹腔处进行了测试。研究结果表明:该方法能够有效优化刀具路径,最大形状偏差从70μm减小到35μm,降低了50%。此外,该方法具有较好的知识可传递性。Aiming at the automatic compensation of tool deflection during milling,a self-optimizing method of tool path for 5-axis CNC machining was proposed.This method obtains the process conditions from the milling material removal simulation,and correlates the calculated cutting conditions with corresponding shape error measurements.Secondly,Support Vector Regression(SVR)based on statistical learning theory is used to predict the shape error generated,and self-optimization and tool path generation are performed.It was applied on a 5-axis CNC machine and tested at two cavities.The research results show that the method can effectively adjust the tool path,and the maximum shape deviation is reduced from 70μm to 35μm,which is 50%lower.In addition,this method has better knowledge transferability.

关 键 词：数控加工 刀具路径规划 铣削 机器学习 计算机辅助制造(CAM) 

分 类 号：TH165[机械工程—机械制造及自动化] TG54[金属学及工艺—金属切削加工及机床]