基于声成像与卷积神经网络的轴承故障诊断方法及其可解释性研究

PDF(1910 KB)

振动与冲击 ›› 2022, Vol. 41 ›› Issue (16) : 224-231.

论文

王冉1，石如玉1，胡升涵1，鲁文波2，胡雄1

作者信息 +

An acoustic fault diagnosis method of rolling bearings based on acoustic imaging and convolutional neural network

WANG Ran1, SHI Ruyu1, HU Shenghan1, LU Wenbo2, HU Xiong1

Author information +

文章历史 +

摘要

常用的振动诊断技术一般采用接触式测量，在测量受限的场合具有一定的局限性。本文提出一种具有非接触测量优势的基于声成像与卷积神经网络的滚动轴承声学故障诊断方法。首先，利用传声器阵列获取滚动轴承辐射的空间声场；然后，用波叠加法进行声成像，重建后的声像能够描述声场的空间分布信息；最后，建立卷积神经网络（CNN），使用不同轴承运行状态下的声像样本对CNN模型进行训练用于故障诊断。同时，针对深度学习模型的诊断结果缺乏可解释性的问题，本文采用梯度加权类激活图（Grad-CAM）算法对卷积神经网络在基于声像的轴承故障诊断中的可解释性进行了研究。轴承实验台的声阵列数据验证了所提方法的有效性及优越性。
关键词：声成像；故障诊断；卷积神经网络；波叠加法；梯度加权类激活图

Abstract

Contact measurement is generally used in the Common vibration diagnosis techniques, which has certain limitations in situations where measurement is limited. In this paper, a rolling bearing acoustic fault diagnosis method based on acoustic imaging and convolutional neural network with the advantage of non-contact measurement was proposed. First, the spatial acoustic field radiated by the rolling bearing is obtained by using microphone array; then, acoustic imaging is performed by wave superposition method, and the reconstructed acoustic image can describe the spatial distribution information of the acoustic field; finally, a convolutional neural network (CNN) is established, which is trained for fault diagnosis using the acoustic image samples of different bearing operating states. Meanwhile, to address the problem of lack of interpretability of diagnostic results of deep learning models, this paper investigates the interpretability of convolutional neural networks in acoustic image-based bearing fault diagnosis using the gradient-weighted class activation map (Grad-CAM) algorithm. The acoustic array data from the bearing experimental bench verifies the effectiveness and superiority of the proposed method.

导出引用

王冉1，石如玉1，胡升涵1，鲁文波2，胡雄1. 基于声成像与卷积神经网络的轴承故障诊断方法及其可解释性研究[J]. 振动与冲击, 2022, 41(16): 224-231

WANG Ran1, SHI Ruyu1, HU Shenghan1, LU Wenbo2, HU Xiong1. An acoustic fault diagnosis method of rolling bearings based on acoustic imaging and convolutional neural network[J]. Journal of Vibration and Shock, 2022, 41(16): 224-231

参考文献

[1] 吴晨芳, 杨世锡, 顾希雯,等. 一种基于改进的LeNet-5模型滚动轴承故障诊断方法研究 [J]. 振动与冲击, 2021, 40(12): 55-61.
WU Chen-fang, YANG Shi-xi, GU Xi-wen, et al. An improved fault diagnosis method of rolling bearings based on LeNet-5 [J]. Vibration and Shock, 2021.40 (12): 55-61
[2] WANG R, LIU F, HOU F, et al. A Non-Contact Fault Diagnosis Method for Rolling Bearings Based on Acoustic Imaging and Convolutional Neural Networks [J]. IEEE Access, 2020 , PP(99):1-1.
[3] WANG D, ZHAO X, KOU L L, et al. A simple and fast guideline for generating enhanced/squared envelope spectra from spectral coherence for bearing fault diagnosis [J]. Mechanical Systems and Signal Processing, 2019, 122(MAY1): 754-768.
[4] HUA L, TAO L, XING W, et al. Application of EEMD and improved frequency band entropy in bearing fault feature extraction [J]. ISA Transactions, 2018, 88:170-185
[5] 谢佳琪, 尤伟, 沈长青, 等. 基于改进卷积深度置信网络的轴承故障诊断研究 [J]. 电子测量与仪器学报, 2020,34(02):36-43
XIE Jia-qi, YOU Wei, SHEN Chang-qing, et al. Bearing fault diagnosis based on improved convolution deep belief network [J]. Journal of Electronic Measurement and Instrumentation, 2020,34(02):36-43
[6] Ao Z , Fang L , Shen C , et al. A New Method of Acoustic Signals Separation for Wayside Fault Diagnosis of Train Bearings[M]. Springer International Publishing, 2015.
[7] C. Li, R.-V. Sanchez, G. Zurita, M. Cerrada, D. Cabrera, R.E. Vásquez, Gearbox fault diagnosis based on deep random forest fusion of acoustic and vibratory signals, Mechanical Systems and Signal Processing, 76-77 (2016) 283-293.
[8] 鲁文波, 蒋伟康. 基于近场声全息的故障诊断方法及系统实现 [J]. 振动与冲击, 2013, 32(005): 48-51.
LU Wen-bo, JIANG Wei-kang. Near-field acoustic holography based fault diagnosis and its system development [J]. Vibration and Shock, 2013, 32(005): 48-51.
[9] WANG R, WANG C, LI J, et al. An intelligent fault diagnosis method of rolling element bearing based on acoustic imaging and gabor wavelet transform; proceedings of the 25th International Congress on Sound and Vibration 2018: Hiroshima Calling, ICSV 2018, July 8, 2018 - July 12, 2018, Hiroshima, Japan, F, 2018 [C]. International Institute of Acoustics and Vibration, IIAV.
[10] YAO Y, WANG H, LI S, et al. End-To-End Convolutional Neural Network Model for Gear Fault Diagnosis Based on Sound Signals [J]. Applied ences, 2018, 8(9): 1584.
[11] KUMAR A, GANDHI C P, ZHOU Y Q, et al. Improved deep convolution neural network (CNN) for the identification of defects in the centrifugal pump using acoustic images [J]. Applied Acoustics, 167.
[12] ZHOU B, KHOSLA A, LAPEDRIZA À, et al. Learning Deep Features for Discriminative Localization [M]. 2016.
[13] RS 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.
[14] 张俊鹏, 杨志勃, 陈雪峰, 等. 卷积神经网络在轴承故障诊断中的可解释性探讨 [J]. 轴承, 2020, (07): 54-60.
ZHANG Jun-peng, YANG Zhi-bo, CHEN Xue-feng,et al. Interpretability discussion on convolutional neural network in bearing fault diagnosis [J]. Bearing, 2020, (07): 54-60.
[15] 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.
[16] KOOPMANN G H, SONG L, FAHNLINE J B. A method for computing acoustic fields based on the principle of wave superposition [J]. The Journal of the Acoustical Society of America, 1989, 86(6): 2433-2428.
[17] Bi C X , Liu Y , Zhang Y B , et al. Extension of sound field separation technique based on the equivalent source method in a sparsity framework[J]. Journal of Sound and Vibration, 2018, 442.
[18] Novati P , Russo M R . A GCV based Arnoldi-Tikhonov regularization method[J]. BIT Numerical Mathematics, 2014, 54(2):501-521.
[19] LIN M, CHEN Q, YAN S. Network In Network [J]. Computer Science, 2013.