EMD和BPNN-GA在微铣刀磨损预测中的应用  

Application of EMD and BPNN-GA in Wear Prediction of Micro Milling Tool

作　　者：王二化[1] 郭伟[1] 赵宇航 刘颉 WANG Er-hua;GUO Wei;ZHAO Yu-hang;LIU Jie(Changzhou City Lab of Intelligent Technology for Advanced Manufacturing Equipment Changzhou College of Informa-tion Technology,Jiangsu Changzhou 213164,China;School of Hydropower and Information Engineering Huazhong University of Science and Technology,Hubei Wuhan 430074,China)

机构地区：[1]常州信息职业技术学院常州市高端制造装备智能化技术重点实验室,江苏常州213164 [2]华中科技大学水电与数字化工程学院,湖北武汉430074

出　　处：《机械设计与制造》2022年第12期137-140,146,共5页Machinery Design & Manufacture

基　　金：国家973项目—难加工航空零件数字化制造的基础研究(2011CB706803);常州市高端制造装备智能化技术重点实验室(CM20183004);江苏省高校“青蓝工程”中青年学术带头人项目资助;常州信息职业技术学院“1+1+1”协同培育工程建设项目。

摘　　要：提高智能制造过程中微铣削刀具状态监测的精度和效率,提出了一种基于经验模态分解(Empirical Mode Decomposition,EMD)和BP神经网络-遗传算法(Back-Propagation Neural Networks-Genetic Algorithm,BPNN-GA)的微铣刀磨损预测方法。在此方法中,首先对微铣削振动信号进行EMD分解,提取各个IMF分量的均值、均方根、峭度、偏态作为微铣刀磨损特征。然后通过相关性分析选择与微铣刀磨损特征密切相关的特征,并选择相互之间相关度最小的几个特征作为微铣刀磨损特征,这样既保证了特征对于研究对象的灵敏度,又保证了特征之间的独立性,不会造成信息冗余。最后利用BPNN-GA模型进行特征分类,实现微铣刀磨损的预测。结果表明,本文提出的微铣刀磨损预测方法能够准确识别各种磨损状态,可以为其它刀具状态监测方法提供必要的理论基础和实践意义。In order to improve the accuracy and efficiency of the condition monitoring of micro milling tools in the intelligent manufacturing process,this paper proposes a wear prediction method based on the empirical mode decomposition(EMD)and BP neural network genetic algorithm(BPNN-GA). In this method,firstly,EMD is used to decompose the vibration signals of micro milling,and the mean value,root mean square,kurtosis and skewness of each IMF component are extracted as the wear characteristics of micro milling tool. Then,the features closely related to the wear characteristics of the micro milling tool are selected through correlation analysis,and several features with the smallest correlation degree are selected as the wear characteristics of the micro milling tool,which not only ensures the sensitivity of the features to the research object,but also guarantees the independence of the features,and does not cause information redundancy. Finally,BPNN-GA model is used for feature classification to realize wear prediction of micro milling tool. The results show that the wear prediction method proposed in this paper can accurately identify all kinds of wear conditions,and can provide certain theoretical basis and practical significance for other tool condition monitoring methods.

关 键 词：微铣刀 刀具磨损 经验模态分解 BP神经网络 遗传算法 

分 类 号：TH16[机械工程—机械制造及自动化]