留言板

尊敬的读者、作者、审稿人, 关于本刊的投稿、审稿、编辑和出版的任何问题, 您可以本页添加留言。我们将尽快给您答复。谢谢您的支持!

姓名
邮箱
手机号码
标题
留言内容
验证码

采用一维卷积神经网络的铣削振动状态识别

郑华林 张冲 何勇

郑华林, 张冲, 何勇. 采用一维卷积神经网络的铣削振动状态识别[J]. 机械科学与技术, 2023, 42(7): 1081-1087. doi: 10.13433/j.cnki.1003-8728.20220059
引用本文: 郑华林, 张冲, 何勇. 采用一维卷积神经网络的铣削振动状态识别[J]. 机械科学与技术, 2023, 42(7): 1081-1087. doi: 10.13433/j.cnki.1003-8728.20220059
ZHENG Hualin, ZHANG Chong, HE Yong. Identifying Milling Vibration State Using One-dimensional Convolutional Neural Network[J]. Mechanical Science and Technology for Aerospace Engineering, 2023, 42(7): 1081-1087. doi: 10.13433/j.cnki.1003-8728.20220059
Citation: ZHENG Hualin, ZHANG Chong, HE Yong. Identifying Milling Vibration State Using One-dimensional Convolutional Neural Network[J]. Mechanical Science and Technology for Aerospace Engineering, 2023, 42(7): 1081-1087. doi: 10.13433/j.cnki.1003-8728.20220059

采用一维卷积神经网络的铣削振动状态识别

doi: 10.13433/j.cnki.1003-8728.20220059
基金项目: 

四川省科技厅重点研发项目 19ZDZX0055

详细信息
    作者简介:

    郑华林(1965-), 教授, 博士, 研究方向为智能制造工艺与装备技术, zhl@swpu.edu.cn

  • 中图分类号: TP18

Identifying Milling Vibration State Using One-dimensional Convolutional Neural Network

  • 摘要: 由于铣削加工中发生颤振会极大地降低工件的加工质量,铣削振动状态的高效与精准辨识一直是颤振研究的热点问题之一。基于LetNet-5经典卷积网络提出一维卷积网络模型,直接对时域铣削力信号进行处理与识别,针对信号量较少与数据不均衡等问题,采用重叠-随机协同采样的方法对数据进行处理。应用T-分布随机邻域嵌入技术可视化模型在训练集上的学习进程并对端到端的学习目标进行验证。对比基于支持向量机与卷积神经网络识别策略,所提方案在测试集上取得了最高的96.17%准确率,识别结果表明: 该方法相较于对比方法过程简单、识别快速且辨识准确率高。
  • 图  1  GCNN-1D的辨识模型

    Figure  1.  Identification model for GCNN-1D

    图  2  数据集划分

    Figure  2.  Data set partitioning

    图  3  铣削振动状态辨识的总体辨识的总体辨识流程

    Figure  3.  The overall identification process

    图  4  铣削实验平台

    Figure  4.  Milling experiment platform

    图  5  训练集在各网络层上的学习效果

    Figure  5.  The learning effect of the training set on each network layer

    图  6  训练集与测试集的准确率迭代图

    Figure  6.  Accuracy iteration graph of the training set and test set

    图  7  测试集混淆矩阵

    Figure  7.  Test set confusion matrix

    图  8  不同模型的ROC曲线

    Figure  8.  ROC curves of different models

    表  1  GCNN-1D网络结构及参数

    Table  1.   GCNN-1D network structure and parameters

    网络层(类型) 输出尺寸 参数量
    Conv1d_1 (32, 200, 32) 192
    Batch_normalization_1 (32, 200, 32) 128
    Relu_1 (32, 200, 32) 0
    Mapooling1d_1 (32, 100, 32) 0
    Conv1d_2 (32, 100, 64) 6 208
    Batch_normalization_2 (32, 100, 64) 256
    Relu_2 (32, 100, 64) 0
    Mapooling1d_2 (32, 50, 64) 0
    Conv1d_3 (32, 50, 88) 16 984
    Batch_normalization_3 (32, 50, 88) 352
    Relu_3 (32, 50, 88) 0
    Mapooling1d_3 (32, 25, 88) 0
    Global_average_pooling1d_1 (32, 88) 0
    Fc1 (32, 3) 267
    总参数量: 24, 387
    可训练参数量: 24, 019
    不可训练参数量: 368
    下载: 导出CSV

    表  2  模型的平均性能指标

    Table  2.   Average performance metrics of the model

    识别方案 训练准确率/% 训练耗时/s 测试准确率/% 测试耗时/s 测试准确率误差
    GCNN-1D 98.62 56.91 96.17 0.011 8 6.22×10-4
    CNN2 96.85 478.82 95.51 0.042 2 1.98×10-4
    SVM1 74.93 0.028 7 65.22 0.004 2 2.21×10-2
    SVM2 89.86 0.086 5 85.71 0.001 7 8.71×10-3
    下载: 导出CSV
  • [1] 卢晓红, 王凤晨, 王华, 等. 铣削过程颤振稳定性分析的研究进展[J]. 振动与冲击, 2016, 35(1): 74-82. https://www.cnki.com.cn/Article/CJFDTOTAL-ZDCJ201601015.htm

    LU X H, WANG F C, WANG H, et al. Review about chatter stability analysis in milling process[J]. Journal of Vibration and Shock, 2016, 35(1): 74-82. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-ZDCJ201601015.htm
    [2] COMAK A, BUDAK E. Modeling dynamics and stability of variable pitch and helix milling tools for development of a design method to maximize chatter stability[J]. Precision Engineering, 2017, 47: 459-468. doi: 10.1016/j.precisioneng.2016.09.021
    [3] CEN L J, MELKOTE S N, CASTLE J, et al. A method for mode coupling chatter detection and suppression in robotic milling[J]. Journal of Manufacturing Science and Engineering, 2018, 140(8): 081015. doi: 10.1115/1.4040161
    [4] 谢锋云, 江炜文, 陈红年, 等. 基于广义BP神经网络的切削颤振识别研究[J]. 振动与冲击, 2018, 37(5): 65-70. https://www.cnki.com.cn/Article/CJFDTOTAL-ZDCJ201805011.htm

    XIE F Y, JIANG W W, CHEN H N, et al. Cutting chatter recognition based on generalized BP neural network[J]. Journal of Vibration and Shock, 2018, 37(5): 65-70. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-ZDCJ201805011.htm
    [5] 王桃章, 王宇, 王宇斐, 等. 深度学习在机器人加工颤振辨识中的应用[J]. 机械科学与技术, 2021, 40(2): 188-192. doi: 10.13433/j.cnki.1003-8728.20200036

    WANG T Z, WANG Y, WANG Y F, et al. Application of deep learning in robot milling chattering identification[J]. Mechanical Science and Technology for Aerospace Engineering, 2021, 40(2): 188-192. (in Chinese) doi: 10.13433/j.cnki.1003-8728.20200036
    [6] GAO H N, SHEN D H, YU L, et al. Identification of cutting chatter through deep learning and classification[J]. International Journal of Simulation Modelling, 2020, 19(4): 667-677. doi: 10.2507/IJSIMM19-4-CO16
    [7] TRAN M Q, LIU M K, TRAN Q V. Milling chatter detection using scalogram and deep convolutional neural network[J]. The International Journal of Advanced Manufacturing Technology, 2020, 107(3): 1505-1516.
    [8] 李欣欣, 周学良. 基于卷积神经网络的深孔镗削加工过程颤振监测研究[J]. 湖北汽车工业学院学报, 2018, 32(4): 49-54. https://www.cnki.com.cn/Article/CJFDTOTAL-HQCG201804012.htm

    LI X X, ZHOU X L. Research on chatter monitoring method for deep-hole boring process based on convolutional neural network[J]. Journal of Hubei University of Automotive Technology, 2018, 32(4): 49-54. (in Chinese) https://www.cnki.com.cn/Article/CJFDTOTAL-HQCG201804012.htm
    [9] JANSSENS O, SLAVKOVIKJ V, VERVISCH B, et al. Convolutional neural network based fault detection for rotating machinery[J]. Journal of Sound and Vibration, 2016, 377: 331-345.
    [10] JI S W, XU W, YANG M, et al. 3D convolutional neural networks for human action recognition[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2013, 35(1): 221-231.
    [11] LECUN Y, BOTTOU L, BENGIO Y, et al. Gradient-based learning applied to document recognition[J]. Proceedings of the IEEE, 1998, 86(11): 2278-2324, doi: 10.1109/5.726791.
    [12] 蔡飞. 基于三维卷积神经网络的新生儿疼痛表情识别[D]. 南京: 南京邮电大学, 2018.

    CAI F. Neonatal pain expression recognition based on 3D convolution neural network[D]. Nanjing: Nanjing University of Posts and Telecommunications, 2018. (in Chinese)
    [13] ZHOU B L, KHOSLA A, LAPEDRIZA A, et al. Learning deep features for discriminative localization[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Las Vegas: IEEE, 2016: 2921-2929.
    [14] RUMELHART D E, HINTON G E, WILLIAMS R J. Learning representations by error propagation[R]. San Diego: University of California, 1985.
    [15] RUMELHART D E, HINTON G E, WILLIAMS R J. Learning representations by back-propagating errors[J]. Nature, 1986, 323(6088): 533-536.
    [16] GLOROT X, BENGIO Y. Understanding the difficulty of training deep feedforward neural networks[C]//Proceedings of the 13th International Conference on Artificial Intelligence and Statistics. Sardinia: PMLR, 2010: 249-256.
  • 加载中
图(8) / 表(2)
计量
  • 文章访问数:  140
  • HTML全文浏览量:  49
  • PDF下载量:  24
  • 被引次数: 0
出版历程
  • 收稿日期:  2021-07-12
  • 刊出日期:  2023-07-25

目录

    /

    返回文章
    返回