基于可解释物理引导空间注意力改进的跨工艺参数立铣刀磨损辨识

doi:10.3901/JME.2024.12.147

机械工程学报 ›› 2024, Vol. 60 ›› Issue (12): 147-157.doi: 10.3901/JME.2024.12.147

• 特邀专栏：可解释可信AI驱动的智能监测与诊断 • 上一篇下一篇

扫码分享

基于可解释物理引导空间注意力改进的跨工艺参数立铣刀磨损辨识

赖旭伟¹, 丁昆¹, 张楷^1,2, 黄锋飞¹, 郑庆^1,2, 李致萱¹, 丁国富^1,2

1. 西南交通大学机械工程学院成都 610031;
2. 西南交通大学轨道交通运维技术与装备四川省重点实验室成都 610031

收稿日期:2023-09-01 修回日期:2024-05-06 出版日期:2024-06-20 发布日期:2024-08-23
作者简介:赖旭伟,男,1994年出生,博士研究生。主要研究方向为刀具磨损监测、可解释深度学习。E-mail:2665395992@qq.com;张楷(通信作者),男,1990年出生,博士,助理教授。主要研究方向为智能故障诊断与寿命预测、复杂装备数字孪生运维技术。E-mail:zhangkai@swjtu.edu.cn
基金资助:
国家重点研发计划(2020YFB1708000)、国家自然科学基金青年基金(52205130)和中央高校基本科研业务费专项资金(2682022CX006)资助项目。

Improved Interpretable-based Physically Guided Spatial Attention for Cross-process Parameters End Milling Cutter Wear Identification

LAI Xuwei¹, DING Kun¹, ZHANG Kai^1,2, HUANG Fengfei¹, ZHENG Qing^1,2, LI Zhixuan¹, DING Guofu^1,2

1. School of Mechanical Engineering, Southwest Jiaotong University, Chengdu 610031;
2. Technology and Equipment of Rail Transit Operation and Maintenance Key Laboratory of Sichuan Province, Southwest Jiaotong University, Chengdu 610031

Received:2023-09-01 Revised:2024-05-06 Online:2024-06-20 Published:2024-08-23

摘要/Abstract

摘要： 深度学习的可解释性不足已成为制约其工业应用的重要问题。刀具磨损状态智能评估方法有着较高的速度、自动化和智能化程度，然而其端到端的模式及提取的特征难以被理解。尤其在跨工艺参数时，其可解释性差、可靠性不足。为此提出一种基于可解释物理引导空间注意力机制改进的跨工艺参数立铣刀磨损辨识方法，首先根据信号的周期不连续特性构建了一种物理引导的空间注意力模块，实现了在跨工艺参数(进给速度、切深)下关键信号片段的自适应捕捉；其次结合最大均值差异及方差约束对特征进行约束，缩小了不同工艺参数下磨损特征的分布差异，同时提升了同类磨损特征的聚集程度。通过实际的刀具切削试验验证，所提方法有效提升了跨工艺参数下的辨识精度；空间注意力权重分布的可视化分析结果表明，所提方法提取出的特征具有明显的物理可解释性。所提方法可为刀具磨损监测的深度辨识模型进行可解释性优化提供参考。

关键词: 可解释深度学习, 刀具磨损, 空间注意力, 物理特性

Abstract: The insufficient interpretability of deep learning has become a critical issue restraining its industrial applications. Intelligent assessment methods for tool wear state exhibit high levels of speed, automation, and intelligence; however, the end-to-end patterns and extracted features are challenging to be understood. Especially when dealing with cross-process parameters, their interpretability is poor, and their reliability is insufficient. Cross-process parameters end milling cutter wear state identification model is proposed to address this based on an improved interpretable-based physically guided spatial attention mechanism. Firstly, a physically guided spatial attention module is constructed based on the periodic discontinuity characteristics of the signals, enabling the adaptive capture of key signal fragments under cross-process parameters such as feed rate and cutting depth. Secondly, constraints are applied to the features using maximum mean discrepancy and variance, reducing the distribution discrepancies of wear features under different process parameters while enhancing the aggregation of similar wear features. Through practical cutting experiments on the tool, the effectiveness of the proposed method in improving the identification accuracy under cross-process parameters is validated. Visual analysis of the spatial attention weight distribution indicates that the proposed method's extracted features possess obvious physical interpretability. The proposed method can provide a reference for the interpretability optimization of the deep learning-based identification model of tool wear monitoring.

Key words: interpretability deep learning, tool wear, spatial attention, physical characteristics

中图分类号:

TP206

赖旭伟, 丁昆, 张楷, 黄锋飞, 郑庆, 李致萱, 丁国富. 基于可解释物理引导空间注意力改进的跨工艺参数立铣刀磨损辨识[J]. 机械工程学报, 2024, 60(12): 147-157.

LAI Xuwei, DING Kun, ZHANG Kai, HUANG Fengfei, ZHENG Qing, LI Zhixuan, DING Guofu. Improved Interpretable-based Physically Guided Spatial Attention for Cross-process Parameters End Milling Cutter Wear Identification[J]. Journal of Mechanical Engineering, 2024, 60(12): 147-157.

参考文献

[1] TRÖBER P，WEISS H A，KOPP T，et al. On the correlation between thermoelectricity and adhesive tool wear during blanking of aluminum sheets[J]. International Journal of Machine Tools and Manufacture，2017，118/119：91-97.
[2] 狄子钧，袁东风，李东阳，等. 基于多尺度-高效通道注意力网络的刀具故障诊断方法[J]. 机械工程学报，2024，60(6)：82-90. DI Zijun，YUAN Dongfeng，LI Dongfang，et al. Tool fault diagnosis method based on multiscale-efficient channel attention network[J]. Journal of Mechanical Engineering，2024，60(6)：82-90.
[3] DOU J，XU C，JIAO S，et al. An unsupervised online monitoring method for tool wear using a sparse auto-encoder[J]. International Journal of Advanced Manufacturing Technology，2020，106：9-12.
[4] CHEN Y，JIN Y，JIRI G. Predicting tool wear with multi-sensor data using deep belief networks[J]. International Journal of Advanced Manufacturing Technology，2018，99：1917-1926.
[5] ZOU Y，DING K，SHI K，et al. Wear identification of end mills based on a feature-weighted convolutional neural network under unbalanced samples[J]. Journal of Manufacturing Processes，2023，89：64-76.
[6] ZHOU Y，ZHI G，CHEN W，et al. A new tool wear condition monitoring method based on deep learning under small samples[J]. Measurement，2022，189：110622.
[7] 史珂铭，邹益胜，刘永志，等.一种不同工艺条件下刀具磨损状态多类域适应迁移辨识方法[J]. 中国机械工程，2022，33(15)：1841-1849. SHI Keming，ZOU Yisheng，LIU Yongzhi，et al. A multi class domain adaptive transfer identification method for tool wear states under different processing conditions[J]. China Mechanical Engineering，2022，33(15)：1841-1849.
[8] HE J，SUN Y，YIN C，et al. Cross-domain adaptation network based on attention mechanism for tool wear prediction[J/OL]. Journal of Intelligent Manufacturing，2022：1-23. https://link.springer.com/article/10.1007/ s10845-022-02005-z#citeas.
[9] GUO H，ZHANG Y，ZHU K. Interpretable deep learning approach for tool wear monitoring in high-speed milling[J]. Computers in Industry，2022，138：103638.
[10] WANG F，ZHAO Z，ZHAI Z，et al. Explainability-driven model improvement for SOH estimation of lithium-ion battery[J]. Reliability Engineering and System Safety，2023，232：109046.
[11] MILLER T. Explanation in artificial intelligence：Insights from the social sciences[J]. Artificial Intelligence：An International Journal，2019，267：1-38.
[12] SELVARAJU R R，COGSWELL M，DAS A，et al. Grad-CAM：Visual explanations from deep networks via gradient-based localization[J]. International Journal of Computer Vision，2020，128(2)：336-359.
[13] BACH S，BINDER A，MONTAVON G，et al. On pixel-wise explanations for nonlinear classifier decisions by layer-wise relevance propagation[J]. PLoS ONE，2015，10(7)：1-46.
[14] CHEN Q，DONG X，TU G，et al. TFN：An interpretable neural network with time-frequency transform embedded for intelligent fault diagnosis[J]. arXiv:2209.01992，2022.
[15] WANG D，CHEN Y，SHEN C，et al. Fully interpretable neural network for locating resonance frequency bands for machine condition monitoring[J]. Mechanical Systems and Signal Processing，2022，168：108673.
[16] LI T，ZHAO Z，SUN C，et al. WaveletKernelNet：An interpretable deep neural network for industrial intelligent diagnosis[J]. IEEE Transactions on Systems，Man，and Cybernetics：Systems，2022，54(4)：2302-2312.
[17] ZHAO Z，LI T，AN B，et al. Model-driven deep unrolling：Towards interpretable deep learning against noise attacks for intelligent fault diagnosis[J]. ISA Transactions，2022，129(B)：644-662.
[18] WANG J，LI Y，ZHAO R，et al. Physics guided neural network for machining tool wear prediction[J]. Journal of Manufacturing Systems，2020，57：298-310.
[19] LI Y，WANG J，HUANG Z，et al. Physics-informed meta learning for machining tool wear prediction[J]. Journal of Manufacturing Systems，2022，62：17-27.
[20] ZHANG W，LI C，PENG G，et al. A deep convolutional neural network with new training methods for bearing fault diagnosis under noisy environment and different working load[J]. Mechanical Systems and Signal Processing，2018，100：439-453.
[21] GRETTON A，BORGWARDT K M，RASCH M，et al. A kernel method for the two-sample-problem[C/CD]// Advances in Neural Information Processing Systems. 2007.
[22] ZOU Y，LIU Y，DENG J，et al. A novel transfer learning method for bearing fault diagnosis under different working conditions[J]. Measurement，2021，171：108767.
[23] MOHAMED A，HASSAN M，SAOUBI R M，et al. Tool condition monitoring for high-performance machining systems-a review[J]. Sensors，2022，22(6)：2206.
[24] ZHANG Y，ZHU K，DUAN X，et al. Tool wear estimation and life prognostics in milling：Model extension and generalization[J]. Mechanical Systems and Signal Processing，2021，155：107617.
[25] GANIN Y，USTINOVA E，AJAKAN H，et al. Domain-adversarial training of neural networks[J]. Advances in Computer Vision and Pattern Recognition，2017，17：189-209.
[26] SUN B，SAENKO K. Deep CORAL：Correlation alignment for deep domain adaptation[J]. Lecture Notes in Computer Science，2016，9915：443-450.
[27] van der MAATEN L. Accelerating t-SNE using tree-based algorithms[J]. Journal of Machine Learning Research，2015，15：3221-3245.

基于可解释物理引导空间注意力改进的跨工艺参数立铣刀磨损辨识

Improved Interpretable-based Physically Guided Spatial Attention for Cross-process Parameters End Milling Cutter Wear Identification

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics

本文评价

[1]	刘岳开, 王天杨, 褚福磊. 基于低延迟可解释性深度学习的复杂旋转机械关键部件知识嵌入与诊断方法研究[J]. 机械工程学报, 2024, 60(12): 107-115.
[2]	王润琼, 宋清华, 彭业振, 刘战强. 基于特征自适应融合和集成学习的高性能铣削刀具状态监测[J]. 机械工程学报, 2024, 60(1): 149-158.
[3]	都建标, 张强, 宗文俊. 硬脆及黑色金属材料的单点金刚石车削加工技术综述[J]. 机械工程学报, 2023, 59(7): 156-175.
[4]	黄景山, 刘国亮, 孙浩, 姚斌, 金林, 陈彬强, 曹新城, 王山城, 许浩, 蔡志钦. C60纳米粒子切削液对15Cr14Co12Mo5Ni2WA齿轮钢切削特性的影响研究[J]. 机械工程学报, 2023, 59(23): 358-371.
[5]	刘洪成, 袁德志, 朱锟鹏. 基于高斯过程潜力模型的刀具磨损预测[J]. 机械工程学报, 2023, 59(17): 310-324.
[6]	段暕, 周宏娣, 刘智勇, 詹小斌, 梁健强, 史铁林. 基于改进PCANet模型的铣刀磨损预测方法研究[J]. 机械工程学报, 2023, 59(1): 278-285.
[7]	吴彤, 李冰锋, 费树岷, 连东辉. 基于改进FCOS算法的架空输电线路防振锤检测^*[J]. 电气工程学报, 2023, 18(1): 143-152.
[8]	滕洪钊, 邓朝晖, 吕黎曙, 谷倩微, 刘涛, 卓荣锦. 多传感器信息融合的加工过程状态监测研究[J]. 机械工程学报, 2022, 58(6): 26-41.
[9]	赵国龙, 夏宏军, 李亮, 信连甲, 何宁. 激光诱导可控氧化辅助微细铣削TiAl金属间化合物的研究[J]. 机械工程学报, 2021, 57(9): 254-263.
[10]	李聪波, 余必胜, 肖溱鸽, 孙鑫, 吕岩. 考虑刀具磨损的数控车削批量加工工艺参数节能优化方法[J]. 机械工程学报, 2021, 57(1): 217-229.
[11]	安华, 王喆, 王国锋, 宋庆月, 刘峰, 钟才川. 复合材料钻削表面粗糙度在线监测与加工参数自适应优化[J]. 机械工程学报, 2020, 56(2): 27-34,42.
[12]	李亚, 黄亦翔, 赵路杰, 刘成良. 基于t分布邻域嵌入与XGBoost的刀具多工况磨损评估[J]. 机械工程学报, 2020, 56(1): 132-140.
[13]	刘佳佳, 姜兴刚, 张德远. 钛合金高速旋转超声椭圆振动侧铣削切屑特征和刀具磨损研究[J]. 机械工程学报, 2019, 55(19): 186-194.
[14]	陈涛, 宋立星, 李素燕, 郭健, 刘献礼. 渐变强化刃刀具硬切削过程切屑流动和刀具磨损研究[J]. 机械工程学报, 2019, 55(19): 195-200.
[15]	郭飞燕, 刘检华, 邹方, 翟雨农, 王仲奇, 李少卓. 数字孪生驱动的装配工艺设计现状及关键实现技术研究[J]. 机械工程学报, 2019, 55(17): 110-132.