基于深度学习与粒子滤波的刀具寿命预测
|
摘要
刀具在加工过程中会受到材料的挤压、摩擦、冲击与腐蚀等因素影响,导致切削刃出现崩刃、磨损等现象.这些现象使得工件尺寸出现偏差,严重时甚至会对机床和人员带来伤害.有效的刀具剩余使用寿命预测可以提高加工效率,保证加工精度,降低加工成本,因此具有重要的研究价值.针对反映刀具磨损程度的趋势性特征自学习提取与刀具剩余使用寿命预测问题,提出了基于深度学习与混合趋势粒子滤波的刀具剩余使用寿命预测方法.使用刀具未发生磨损的信号特征训练降噪自编码器,然后将刀具各磨损阶段下的信号特征输入训练好的降噪自编码器中,提取其重构误差作为单调性特征,为了解决样本数量不足带来的过拟合的问题,对原始样本进行了加噪处理.考虑到传统粒子滤波算法进行刀具剩余使用寿命预测的过程中无法自适应调整状态方程,提出混合趋势粒子滤波算法来实现刀具剩余使用寿命预测.采集刀具全寿命周期的切削力信号并进行处理与分析,分析结果证明了所提方法能够有效实现反映刀具磨损的趋势性特征自提取,该特征提取方法可以减少人为因素的影响,降低训练成本,同时,相比于传统粒子滤波,混合趋势粒子滤波算法对刀具剩余使用寿命预测精度更加准确可靠.
Tools are affected by extrusion,friction,impact,and corrosion during machining,and these result in chipping and wearing of tools. These can cause deviations in the workpiece and cause damage to machines and personnel. Effective prediction of tools' remaining useful life has important research value,as it can greatly improve the quality of workpiece,guarantee processing accuracy,and reduce processing costs. To realize self-extraction of features for tool wear and predict a tools' remaining useful life,a method based on deep learning and hybrid trend particle filtering is proposed in this study. A neural network was trained using cutting force signal of a normal tool without wear,and reconstruction error was extracted as a monotonic feature. To solve the problem of over-fitting caused by insufficient sample size,noise was added to the original sample. To overcome traditional particle filtering algorithms' inability to adaptively adjust state equation during a tools' remaining useful life prediction process,a hybrid trend particle filter algorithm is proposed to realize the tool life prediction. Cutting force signals of a tools' life cycle are collected for analysis. The experimental results prove that the proposed method can effectively achieve trend feature self-extraction. Moreover,the method can also effectively reduce the influence of human factors and reduce training cost. Furthermore,compared with the traditional particle filter,the hybrid trend particle filter algorithm is more accurate and reliable in predicting tools' remaining useful life.
Tools are affected by extrusion,friction,impact,and corrosion during machining,and these result in chipping and wearing of tools. These can cause deviations in the workpiece and cause damage to machines and personnel. Effective prediction of tools' remaining useful life has important research value,as it can greatly improve the quality of workpiece,guarantee processing accuracy,and reduce processing costs. To realize self-extraction of features for tool wear and predict a tools' remaining useful life,a method based on deep learning and hybrid trend particle filtering is proposed in this study. A neural network was trained using cutting force signal of a normal tool without wear,and reconstruction error was extracted as a monotonic feature. To solve the problem of over-fitting caused by insufficient sample size,noise was added to the original sample. To overcome traditional particle filtering algorithms' inability to adaptively adjust state equation during a tools' remaining useful life prediction process,a hybrid trend particle filter algorithm is proposed to realize the tool life prediction. Cutting force signals of a tools' life cycle are collected for analysis. The experimental results prove that the proposed method can effectively achieve trend feature self-extraction. Moreover,the method can also effectively reduce the influence of human factors and reduce training cost. Furthermore,compared with the traditional particle filter,the hybrid trend particle filter algorithm is more accurate and reliable in predicting tools' remaining useful life.
引文
[1]Wang J,Wang P,Gao R X.Enhanced particle filter for tool wear prediction[J].Journal of Manufacturing Systems,2015,36:35-45.
[2]吴德林,周云飞.高速铣削刀具磨损寿命实验及建模研究[J].制造技术与机床,2008(11):84-87.Wu Delin,Zhou Yunfei.Modeling and experimental study on tool wear life in high-speed milling[J].Manufacturing Technology&Machine Tool,2008(11):84-87(in Chinese).
[3]孙惠斌,牛伟龙,王俊阳.基于希尔伯特黄变换的刀具磨损特征提取[J].振动与冲击,2015,34(4):158-164.Sun Huibin,Niu Weilong,Wang Junyang.Tool wear feature extraction based on Hilbert-Huang transformation[J].Journal of Vibration and Shock,2015,34(4):158-164(in Chinese).
[4]Wang P,Gao R X.Adaptive resampling-based particle filtering for tool life prediction[J].Journal of Manufacturing Systems,2015,37:528-534.
[5]Jouin M,Gouriveau R,Hissel D,et al.Particle filterbased prognostics:Review,discussion and perspectives[J].Mechanical Systems&Signal Processing,2016,72/73:2-31.
[6]刘锐,王玫,陈勇.铣刀磨损量监测和剩余寿命预测方法研究[J].现代制造工程,2010(6):102-105.Liu Rui,Wang Mei,Chen Yong.A methodology for on-line tool wear monitoring and predicting the remaining useful life of the cutting tool in face milling[J].Modern Manufacturing Engineering,2010(6):102-105(in Chinese).
[7]关山,闫丽红,彭昶.LS-SVM回归算法在刀具磨损量预测中的应用[J].中国机械工程,2015,26(2):217-222.Guan Shan,Yan Lihong,Peng Chang.Application of regression algorithm of LS-SVM in tool wear prediction[J].China Mechanical Engineering,2015,26(2):217-222(in Chinese).
[8]王晓强,张云,周华民,等.基于隐马尔可夫模型的刀具磨损连续监测[J].组合机床与自动化加工技术,2016(10):87-90.Wang Xiaoqiang,Zhang Yun,Zhou Huamin,et al.Continuous tool wear monitoring based on hidden Markov model[J].Modular Machine Tool&Automatic Manufacturing,2016(10):87-90(in Chinese).
[9]Wang Guofeng,Qian Lei,Guo Zhiwei,et al.Continuous tool wear prediction based on Gaussian mixture regression model[J].International Journal of Advanced Manufacturing Technology,2013,66(9/10/11/12):1921-1929.
[10]王国锋,李志猛,董毅.刀具状态智能监测研究进展[J].航空制造技术,2018,61(6):16-23.Wang Guofeng,Li Zhimeng,Dong Yi.Recent advances in intelligent monitoring of cutting tool condition[J].Aeronautical Manufacturing Technology,2018,61(6):16-23(in Chinese).
[11]李巍华,单外平,曾雪琼.基于深度信念网络的轴承故障分类识别[J].振动工程学报,2016,29(2):340-347.Li Weihua,Shan Waiping,Zeng Xueqiong.Bearing fault identification based on deep belief network[J].Journal of Vibration Engineering,2016,29(2):340-347(in Chinese).
[12]朱煜奇,黄双喜,杨天祺,等.基于栈式降噪自编码的故障诊断[J].制造业自动化,2017,39(3):152-156.Zhu Yuqi,Huang Shuangxi,Yang Tianqi,et al.Fault diagnosis based on stacked denoising autoencoder[J].Manufacturing Automation,2017,39(3):152-156(in Chinese).
[13]陈仁祥,杨星,杨黎霞,等.栈式稀疏加噪自编码深度神经网络的滚动轴承损伤程度诊断[J].振动与冲击,2017,36(21):125-131.Chen Renxiang,Yang Xing,Yang Lixia,et al.Fault severity diagnosis method for rolling bearings based on a stacked sparse denoising auto-encoder[J].Journal of Vibration and Shock,2017,36(21):125-131(in Chinese).
[14]单外平,曾雪琼.基于深度信念网络的信号重构与轴承故障识别[J].电子设计工程,2016,24(4):67-71.Shan Waiping,Zeng Xueqiong.Signal reconstruction and bearing fault identification based on deep belief network[J].Electronic Design Engineering,2016,24(4):67-71(in Chinese).
[15]林杨,高思煜,刘同舜,等.基于深度学习的高速铣削刀具磨损状态预测方法[J].机械与电子,2017,35(7):12-17.Lin Yang,Gao Siyu,Liu Tongshun,et al.A deep learning-based method for tool wear state prediction in high speed milling[J].Machinery&Electronics,2017,35(7):12-17(in Chinese).
[16]张绍辉.基于多路稀疏自编码的轴承状态动态监测[J].振动与冲击,2016,35(19):125-131.Zhang Shaohui.Bearing condition dynamic monitoring based on multi-way sparse autocoder[J].Journal of Vibration and Shock,2016,35(19):125-131(in Chinese).
[17]孙磊,贾云献,蔡丽影,等.粒子滤波参数估计方法在齿轮箱剩余寿命预测中的应用研究[J].振动与冲击,2013,32(6):6-12.Sun Lei,Jia Yunxian,Cai Liying,et al.Residual useful life prediction of gearbox based on particle filtering parameter estimation method[J].Journal of Vibration and Shock,2013,32(6):6-12(in Chinese).
[18]Vincent P,Larochelle H,Bengio Y,et al.Extracting and composing robust features with denoising autoencoders[C]//International Conference on Machine Learning.Montreal,Canada,2008:1096-1103.
[19]陈仁祥,黄鑫,杨黎霞,等.加噪样本扩展深度稀疏自编码神经网络的滚动轴承寿命阶段识别[J].振动工程学报,2017,30(5):874-882.Chen Renxiang,Huang Xin,Yang Lixia,et al.Bearing life state recognition using deep sparse auto-encoder neural network with noise adding sample expansion[J].Journal of Vibration Engineering,2017,30(5):874-882(in Chinese).
[20]张西宁,向宙,夏心锐,等.堆叠自编码网络性能优化及其在滚动轴承故障诊断中的应用[J].西安交通大学学报,2018,52(10):49-56,87.Zhang Xining,Xiang Zhou,Xia Xinrui,et al.Optimization of staking auto-encoder and its application in bearing fault diagnosis[J].Journal of Xi’an Jiaotong University,2018,52(10):49-56,87(in Chinese).
[2]吴德林,周云飞.高速铣削刀具磨损寿命实验及建模研究[J].制造技术与机床,2008(11):84-87.Wu Delin,Zhou Yunfei.Modeling and experimental study on tool wear life in high-speed milling[J].Manufacturing Technology&Machine Tool,2008(11):84-87(in Chinese).
[3]孙惠斌,牛伟龙,王俊阳.基于希尔伯特黄变换的刀具磨损特征提取[J].振动与冲击,2015,34(4):158-164.Sun Huibin,Niu Weilong,Wang Junyang.Tool wear feature extraction based on Hilbert-Huang transformation[J].Journal of Vibration and Shock,2015,34(4):158-164(in Chinese).
[4]Wang P,Gao R X.Adaptive resampling-based particle filtering for tool life prediction[J].Journal of Manufacturing Systems,2015,37:528-534.
[5]Jouin M,Gouriveau R,Hissel D,et al.Particle filterbased prognostics:Review,discussion and perspectives[J].Mechanical Systems&Signal Processing,2016,72/73:2-31.
[6]刘锐,王玫,陈勇.铣刀磨损量监测和剩余寿命预测方法研究[J].现代制造工程,2010(6):102-105.Liu Rui,Wang Mei,Chen Yong.A methodology for on-line tool wear monitoring and predicting the remaining useful life of the cutting tool in face milling[J].Modern Manufacturing Engineering,2010(6):102-105(in Chinese).
[7]关山,闫丽红,彭昶.LS-SVM回归算法在刀具磨损量预测中的应用[J].中国机械工程,2015,26(2):217-222.Guan Shan,Yan Lihong,Peng Chang.Application of regression algorithm of LS-SVM in tool wear prediction[J].China Mechanical Engineering,2015,26(2):217-222(in Chinese).
[8]王晓强,张云,周华民,等.基于隐马尔可夫模型的刀具磨损连续监测[J].组合机床与自动化加工技术,2016(10):87-90.Wang Xiaoqiang,Zhang Yun,Zhou Huamin,et al.Continuous tool wear monitoring based on hidden Markov model[J].Modular Machine Tool&Automatic Manufacturing,2016(10):87-90(in Chinese).
[9]Wang Guofeng,Qian Lei,Guo Zhiwei,et al.Continuous tool wear prediction based on Gaussian mixture regression model[J].International Journal of Advanced Manufacturing Technology,2013,66(9/10/11/12):1921-1929.
[10]王国锋,李志猛,董毅.刀具状态智能监测研究进展[J].航空制造技术,2018,61(6):16-23.Wang Guofeng,Li Zhimeng,Dong Yi.Recent advances in intelligent monitoring of cutting tool condition[J].Aeronautical Manufacturing Technology,2018,61(6):16-23(in Chinese).
[11]李巍华,单外平,曾雪琼.基于深度信念网络的轴承故障分类识别[J].振动工程学报,2016,29(2):340-347.Li Weihua,Shan Waiping,Zeng Xueqiong.Bearing fault identification based on deep belief network[J].Journal of Vibration Engineering,2016,29(2):340-347(in Chinese).
[12]朱煜奇,黄双喜,杨天祺,等.基于栈式降噪自编码的故障诊断[J].制造业自动化,2017,39(3):152-156.Zhu Yuqi,Huang Shuangxi,Yang Tianqi,et al.Fault diagnosis based on stacked denoising autoencoder[J].Manufacturing Automation,2017,39(3):152-156(in Chinese).
[13]陈仁祥,杨星,杨黎霞,等.栈式稀疏加噪自编码深度神经网络的滚动轴承损伤程度诊断[J].振动与冲击,2017,36(21):125-131.Chen Renxiang,Yang Xing,Yang Lixia,et al.Fault severity diagnosis method for rolling bearings based on a stacked sparse denoising auto-encoder[J].Journal of Vibration and Shock,2017,36(21):125-131(in Chinese).
[14]单外平,曾雪琼.基于深度信念网络的信号重构与轴承故障识别[J].电子设计工程,2016,24(4):67-71.Shan Waiping,Zeng Xueqiong.Signal reconstruction and bearing fault identification based on deep belief network[J].Electronic Design Engineering,2016,24(4):67-71(in Chinese).
[15]林杨,高思煜,刘同舜,等.基于深度学习的高速铣削刀具磨损状态预测方法[J].机械与电子,2017,35(7):12-17.Lin Yang,Gao Siyu,Liu Tongshun,et al.A deep learning-based method for tool wear state prediction in high speed milling[J].Machinery&Electronics,2017,35(7):12-17(in Chinese).
[16]张绍辉.基于多路稀疏自编码的轴承状态动态监测[J].振动与冲击,2016,35(19):125-131.Zhang Shaohui.Bearing condition dynamic monitoring based on multi-way sparse autocoder[J].Journal of Vibration and Shock,2016,35(19):125-131(in Chinese).
[17]孙磊,贾云献,蔡丽影,等.粒子滤波参数估计方法在齿轮箱剩余寿命预测中的应用研究[J].振动与冲击,2013,32(6):6-12.Sun Lei,Jia Yunxian,Cai Liying,et al.Residual useful life prediction of gearbox based on particle filtering parameter estimation method[J].Journal of Vibration and Shock,2013,32(6):6-12(in Chinese).
[18]Vincent P,Larochelle H,Bengio Y,et al.Extracting and composing robust features with denoising autoencoders[C]//International Conference on Machine Learning.Montreal,Canada,2008:1096-1103.
[19]陈仁祥,黄鑫,杨黎霞,等.加噪样本扩展深度稀疏自编码神经网络的滚动轴承寿命阶段识别[J].振动工程学报,2017,30(5):874-882.Chen Renxiang,Huang Xin,Yang Lixia,et al.Bearing life state recognition using deep sparse auto-encoder neural network with noise adding sample expansion[J].Journal of Vibration Engineering,2017,30(5):874-882(in Chinese).
[20]张西宁,向宙,夏心锐,等.堆叠自编码网络性能优化及其在滚动轴承故障诊断中的应用[J].西安交通大学学报,2018,52(10):49-56,87.Zhang Xining,Xiang Zhou,Xia Xinrui,et al.Optimization of staking auto-encoder and its application in bearing fault diagnosis[J].Journal of Xi’an Jiaotong University,2018,52(10):49-56,87(in Chinese).