基于贝叶斯网络的切削加工表面粗糙度在线监测方法  

Surface roughness monitoring method based on Bayesian network models

作　　者：王明微[1] 周竞涛[1] 敬石开[2] 田国良[1] 

机构地区：[1]西北工业大学现代设计与集成制造技术教育部重点实验室,陕西西安710072 [2]北京理工大学机械与车辆学院,北京100081

出　　处：《计算机集成制造系统》2014年第12期3075-3081,共7页Computer Integrated Manufacturing Systems

基　　金：国家自然科学基金资助项目(51205321;61104169);国家科技支撑计划资助项目(2014BAF07B0);陕西省自然科学基金资助项目(2014JM9367);中央高校基本科研业务费专项资金资助项目(3102014KYJD038)~~

摘　　要：针对加工过程中各种因素对表面粗糙度影响的不确定性,提出一种基于贝叶斯网络的表面粗糙度监测模型。直接从切削力和工件振动的传感器监测信号提取时域和频域能量特征,基于贝叶斯网络学习过程挖掘出表面粗糙度状态与信号特征的关联关系,从而根据粗糙度值域的概率分布得到监测结果。通过铣削加工过程的粗糙度监测实验验证了所提模型的有效性。Aiming at the uncertainty of various factors on surface roughness, a surface roughness prediction modelbased on Bayesian network was proposed. The energy features of time domain and frequency domain were extracteddirectly from cutting force and workpiece vibration. The association relationship between surface roughness and sensor signal features were mined with Bayesian network learning process, and the monitoring result was obtained according to the probability distribution of roughness range. The effectiveness of proposed method was verified by theroughness experiment of milling process.

关 键 词：表面粗糙度监测 贝叶斯网络 传感器信号特征 机器学习 切削加工 

分 类 号：TP182[自动化与计算机技术—控制理论与控制工程]