为了准确监测铣削加工过程中刀具磨损程度,提出了一种基于双向门控循环神经网络融合注意力机制(ABiGRU)的刀具磨损监测模型。在该监测模型中,通过对振动、力和声发射传感器采集到的时序数据进行时域、频域和时频域分析,使用spearman相关系数提取与后刀面平均磨损量强相关的20维特征。引入ELU激活函数来优化BiGRU网络,解决梯度消失问题;利用内部注意力机制提升模型对于重要特征信息的捕捉能力,快速实现从特征到刀具磨损值的映射。通过与RNN、LSTM、GRU、BiLSTM和BiGRU进行的对比分析,结果表明:该模型能够准确地表征刀具磨损程度,并使模型的精度和效率得到了较大的提高。
Abstract
In order to accurately monitor tool wear in the milling process, a tool wear monitoring model based on attention and bidirectional gated recurrent unit (ABiGRU) was proposed. In the monitoring model, firstly, the time series data collected by vibration, force, and acoustic emission sensors were analyzed in the time domain, frequency domain, and time-frequency domain, and spearman coefficient was used to extract 20-dimensional features strongly correlated with the average rear tool surface wear. Secondly, the ELU activation function was introduced to optimize the BiGRU network and solve the gradient disappearance problem, in the meantime, the internal attention mechanism was used to improve the models ability to capture important feature information and quickly realize the mapping from feature to tool wear value. Finally, compared with RNN, LSTM, GRU, BiLSTM, and BiGRU, the results shown that the model could not only accurately characterize the tool wear degree, but also the accuracy and efficiency is improved greatly.
关键词
计量学;刀具磨损监测;双向门控循环神经网络;注意力机制;特征提取 /
ELU激活函数 /
梯度消失
Key words
metrology /
tool wear monitoring /
bidirectional gated recurrent unit /
attention mechanism /
feature extraction /
ELU activation function /
gradient disappearance
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
［1］何志坚, 周志雄, 黄向明. 基于变分模态分解的关联维数及相关向量机的刀具磨损状态监测［J］. 计量学报, 2018, 39(2): 182-186.
He Z J, Zhou Z X. Tool Wear State Monitoring Based on Variational Mode Decomposition and Correlation Dimension and Relevance Vector Machine［J］. Acta Metrologica Sinica, 2018, 39(2): 182-186.
［2］陈剑, 蔡坤奇, 陶善勇等. 基于IITD模糊熵与随机森林的滚动轴承故障诊断方法［J］. 计量学报, 2021, 42(6): 774-779.
Chen J, Cai K Q, Tao S Y, et al. Fault Diagnosis Method of Rolling Bearing Based on IITD Fuzzy Entropy and Random Forest［J］. Acta Metrologica Sinica, 2021, 42(6): 774-779.
［3］Yu J, Liang S, Tang D, et al. A weighted hidden Markov model approach for continuous-state tool wear monitoring and tool life prediction［J］. The International Journal of Advanced Manufacturing Technology, 2017, 91(1): 201-211.
［4］程菲, 董景彦. 基于时间序列数据和支持向量机的纳米加工AFM刀尖损伤监测［J］. 计量学报, 2019, 40(4): 647-654.
Chen F, Dong J Y. Nano-machining AFM Tip Wear Monitoring Based on Time Series Data and Support Vector Machine［J］. Acta Metrologica Sinica, 2019, 40(4): 647-654.
［5］刘惠, 刘振宇, 郏维强, 等. 深度学习在装备剩余使用寿命预测技术中的研究现状与挑战［J］. 计算机集成制造系统, 2021, 27(1): 34-52.
Liu H, Liu Z Y, Jia W Q, et al. Current research and challenges of deep learning for equipment remaining useful life prediction［J］. Computer Integrated Manufacturing System, 2021, 27(1): 34-52.
［6］Schmidhuber J. A local learning algorithm for dynamic feedforward and recurrent networks［J］. Connection Science, 1989, 1(4): 403-412.
［7］李锋, 陈勇, 王家序, 等. 基于强化学习单元匹配循环神经网络的滚动轴承状态趋势预测［J］. 计算机集成制造系统, 2020, 26(8): 2050-2059.
Li F, Chen Y, Wang J X, et al. State trend prediction of rolling bearing based on reinforcement learning unit matching recurrent neural network［J］. Computer Integrated Manufacturing System, 2020, 26(8): 2050-2059.
［8］Zheng S, Ristovski K, Farahat A, et al. Long short-term memory network for remaining useful life estimation［C］//2017 IEEE international conference on prognostics and health management (ICPHM). IEEE, 2017: 88-95.
［9］钟百鸿, 王琳, 钟诗胜. 基于综合灰关联序模型的残差门控循环神经网络位标器零部件选配［J］. 中国机械工程, 2021, 32(3): 314-320, 356.
Zhong B H, Wang L, Zhong S S. Selective Assembly for Coordinator Parts by RNGRU Based on Comprehensive Grey Relational Order Model［J］. Chinese Journal of Mechanical Engineering, 2021, 32(3): 314-320,356.
［10］何彦, 凌俊杰, 王禹林, 等. 基于长短时记忆卷积神经网络的刀具磨损在线监测模型［J］. 中国机械工程, 2020, 31(16): 1959-1967.
He Y, Lin J J, Wang Y L, et al. In-process Tool Wear Monitoring Model Based on LSTM-CNN［J］. Chinese Journal of Mechanical Engineering, 2020, 31(16): 1959-1967.
［11］Wang J, Yan J, Li C, et al. Deep heterogeneous GRU model for predictive analytics in smart manufacturing: Application to tool wear prediction［J］. Computers in Industry, 2019, 111: 1-14.
［12］吴凤和, 钟浩, 章钦, 等. 基于卷积门控循环神经网络的刀具磨损状态监测［J］. 计量学报, 2021, 42(8): 1034-1040.
Wu F H, Zhong H, Zhang Q, et al. Tool Wear Condition Monitoring Based on Convolution-gated Recurrent Neural Network［J］. Acta Metrologica Sinica, 2021, 42(8): 1034-1040.
［13］王佳铖, 鲍劲松, 刘天元, 等. 基于工件注意力的车间作业行为在线识别方法［J］. 计算机集成制造系统, 2021, 27(4): 1099-1107.
Wang J C, Bao J S, Liu T Y, et al. Online method for worker operation recognition based on attention of workpiece［J］. Computer Integrated Manufacturing System, 2021, 27(4): 1099-1107.
［14］王金甲, 周雅倩, 郝智. 基于注意力模型的多传感器人类活动识别［J］. 计量学报, 2019, 40(6): 958-969.
Wang J J, Zhou Y Q, Hao Z. Multi-sensor Human Activity Recognition Based on Attention Model［J］. Acta Metrologica Sinica, 2019, 40(6): 958-969.
［15］郭保苏, 韩天杰,张宇, 等. 基于带阈值模块一维残差网络的刀具磨损监测方法［J］. 计量学报, 2022, 43(4): 501-506.
Guo B S, Han T J, Zhang Y, et al. Tool Wear Monitoring Method Based on One-Dimensional Residual Network with Threshold Module［J］. Acta Metrologica Sinica, 2022, 43(4): 501-506.
［16］Agogino A, Goebel K. Mill Data Set［EB/OL］. http: //tiarc. nasa. gov/project/prognostic-data-repository.
基金
国家科技重大专项(2018ZX04027001);教育部人文社科项目(20YJCZH150);湖北省高等学校优秀中青年科技创新团队计划(T2020018);汽车动力传动与电子控制湖北省重点实验室基金(ZDK1201703);湖北汽车工业学院博士基金(BK201905)
PDF(86019 KB)
