基于注意力机制-Inception-CNN模型的滚动轴承故障分类

PDF(1733 KB)

振动与冲击 ›› 2020, Vol. 39 ›› Issue (19) : 84-93.

论文

朱浩1，宁芊1,2，雷印杰1，陈炳才3，严华1

作者信息 +

Fault classification of rolling bearing based on attention mechanism-inception-CNN Model

ZHU Hao1, NING Qian1,2, LEI Yinjie1, CHEN Bingcai3, YAN Hua1

Author information +

文章历史 +

摘要

针对传统机器学习方法需要大量专家知识和高昂经济成本，研究了一种基于注意力机制和Inception网络结构的卷积神经网络。其注意力机制是对数据的不同特征维度赋予不同的权重，抽取出更加关键和重要的信息，使模型做出更加准确的判断。其Inception网络结构则是拓宽网络的宽度并增加网络对卷积核尺度的适应性，以提取到更加丰富的特征。为了提高模型的泛化能力，在每个卷积层和全连接层后又添加了一个DropBlock层。最后结果显示该模型不仅在同负载的情况下获得很高的滚动轴承故障分类准确率和稳定性，并且在不同负载情况、不同规模的滚动轴承数据集上依旧能保持高的准确率与稳定性。

Abstract

Aiming at traditional machine learning method requiring a lot of expert knowledge and high economic costs, a convolutional neural network (CNN) based on attention mechanism and Inception network structure was proposed. Its attention mechanism was used to assign different weights to different feature dimensions of data, extract more critical and important information, and make the model do more accurate judgments. Its Inception network structure was used to broaden the network width, increase the network’s adaptability to the convolution core scale, and extract more abundant features. In order to improve the generalization ability of the model, a DropBlock layer was added to follow each convolution layer and full connection layer. Finally, the results showed that the proposed model can not only achieve higher fault classification accuracy and stability under the same load, but also maintain higher accuracy and stability under different load conditions and rolling bearing data sets with different scales.

导出引用

朱浩1，宁芊1,2，雷印杰1，陈炳才3，严华1. 基于注意力机制-Inception-CNN模型的滚动轴承故障分类[J]. 振动与冲击, 2020, 39(19): 84-93

ZHU Hao1, NING Qian1,2, LEI Yinjie1, CHEN Bingcai3, YAN Hua1. Fault classification of rolling bearing based on attention mechanism-inception-CNN Model[J]. Journal of Vibration and Shock, 2020, 39(19): 84-93

参考文献

[1]HE Q, WANG J, HU F,et al. Wayside acoustic diagnosis of defective train bearings based on signal resampling and information enhancement[J]. Journal of Sound and Vibration, 2013,332(21):5635-5649.
[2]WANG X, ZHENG Y, ZHAO Z,et al. Bearing fault diag-nosis based on statistical locally linear embedding[J].Sensors,2015,15(7):16225–16247.
[3]Lee W, Park C G. Double fault detection of cone-shaped redundant IMUs using wavelet transformationand EPSA[J]. Sensors,2014,14(2):3428–3444.
[4]Tipping M E, Bishop C M. Probabilistic principal component analysis[J]. Journal of the Royal Statistical Society,2010, 61(3):611-622.
[5]Pandya D H, Upadhyay S H, Harsha S P. Fault diagnosis of rolling element bearing with intrinsic mode function of acoustic emission data using APF-KNN[J]. Expert Systems with Applications,2013,40(10):4137-4145.
[6]Santos P, Villa L F, Reñones A,et al. An SVM-based solution for fault detection in wind turbines[J]. Sensors,2015,15(3): 5627–5648.
[7]Hinton G E, Salakhutdinov R R. Reducing the dimension-ality of data with neural networks[J]. Science,2006,313(5786)：504-507.
[8]孙文珺，邵思羽，严如强. 基于稀疏自动编码深度神经
网络的感应电动机故障诊断[J]. 机械工程学报，2016，
52(9)：65-71.
SUN Wen-jun，SHAO Si-yu，YAN Ru-qiang. Induction
motor fault diagnosis based on sparse auto-encoder deep
neural network [J]. Journal of Mechanical Engineering，
2016，52(9) ：65-71.
[9]Kong Q, Cui G, Yeo S S,et al. DBN wavelet transform denoising method in soybean straw composition based on near-infrared rapid detection[J]. Journal of Real-Time Image Processing,2016,13(3):1-14.
[10]雷亚国，贾峰，周昕，等.基于深度学习理论的机械装备大数据健康监测方法[J].机械工程学报，2015，51（21）：49-56.
LEI Ya-guo, JIA Feng, ZHOU Xin,et al. Large data health monitoring method for mechanical equipment based on deep learning theory [J]. Journal of Mechanical Engineering,2015,51 (21):49-56.
[11]Szegedy C, LIU W, JIA Y Q,et al. Going deeper with convolutions[C]// IEEE Conference on Computer Vision and Pattern Recognition(CVPR). Boston,2015:1-9.
[12]CHEN L, ZHANG H, XIAO J,et al. SCA-CNN: Spatial and channel-wise attention in convolutional networks for image captioning[C]// IEEE Conference on Computer Vision and Pattern Recognition(CVPR). Honolulu,2017:6298–6306.
[13]LIN M, CHEN Q, YAN S. Network in network[C]// International Conference on Learning Representations (ICLR). Banff,2014.
[14]Ghiasi G, Lin T Y, Quoc V L. DropBlock: A regularization method for convolutional networks[C]// neural information processing systems conference(NIPS). Montreal,2018.
[15]ZHANG W, PENG G, LI C,et al. A new deep learning model for fault diagnosis with good anti-noise and domain adaptation ability on raw vibration signals[J]. Sensors,2017, 17(2):425.
[16]Srivastava N, Hinton G, Krizhevsky A,et al. Dropout: A simple way to prevent neural networks from overfitting[J]. The Journal of Machine Learning Research,2014,15(1): 1929-1958.
[17]Kingma D, Ba J. Adam: A method for stochastic optimization[C]// International Conference on Learning Representations (ICLR). San Diego,2015.
[18]Lou X, Loparo K A. Bearing fault diagnosis based on wavelet transform and fuzzy inference[J]. Mechanical Systems and Signal Processing,2004,18(5):1077-1095.
[19]Keskar N S, Mudigere D, Nocedal J,et al. On large-batch training for deep learning: Generalization gap and sharp minima[C]// International Conference on Learning Representations (ICLR). Toulon,2017.
[20]Glorot X, Bengio Y. Understanding the difficulty of training deep feedforward neural networks[J]. Journal of Machine Learning Research,2010,9:249-256.
[21]Ince T, Kiranyaz S, Eren L,et al. Real-time motor fault detection by 1D convolutional neural networks[J]. IEEE Transactions on Industrial Electronics,2016,63(11):7067–7075.
[22]Abdeljaber O, Avci O, Kiranyaz S,et al. Real-time vibrat-ion-based structural damage detection using one-dimensional convolutional neural networks[J]. Journal of Sound & Vibra-tion,2017,388:154-170.
[23]QIU H, LEE J, LIN J,et al. Wavelet filter-based weak signature detection method and its application on rolling element bearing prognostics[J]. Journal of Sound & Vibration,2006, 289(4):1066-1090.
[24] Nectoux P, Gouriveau R, Medjaher K, et al. PRONOSTIA: An experimental platform for bearings accelerated degradation tests[C]//IEEE International Conference on Prognostics and Health Management. Denver,2012.