栈式稀疏加噪自编码深度神经网络的滚动轴承损伤程度诊断

PDF(1393 KB)

振动与冲击 ›› 2017, Vol. 36 ›› Issue (21) : 125-131.

论文

陈仁祥1,2，杨星1，杨黎霞3，王家序2，徐向阳1，陈思杨1

作者信息 +

Fault severity diagnosis method forrolling bearingbased on stacked sparse denoising auto-encoder

CHEN Ren-xiang1,2，YANG Xing1，YANG Li-Xia1，WANG Jia-xu2，XUXiang-yang1，CHEN Si-yang1

Author information +

文章历史 +

摘要

针对滚动轴承损伤程度的特征自学习提取与智能诊断问题，提出栈式稀疏加噪自编码深度神经网络的滚动轴承损伤程度诊断方法。滚动轴承损伤特征受到工况、环境噪声等干扰，浅层自编码网络对损伤特征的自学习、提取能力不足。为此，论文将稀疏项限制和加噪编码融入自编码网络，同时将自编码网络堆栈并添加分类层，构建出栈式稀疏加噪自编码深度神经网络，进行轴承损伤特征非监督自动提取与损伤程度智能诊断。稀疏项限制和深度神经网络的构建提高了特征学习能力，加噪编码的融入改善了网络的鲁棒性。所构建深度神经网络通过多层无监督逐层自学习和有监督微调，完成损伤特征自动提取与表达，并实现了损伤程度智能诊断。不同工况下轴承损伤程度诊断的实验验证证明了所提方法的可行性和有效性。

Abstract

Aiming at solving the problems of fault severity feature self-taught learning and intelligent diagnosis for rolling bearing, a fault severity diagnosis method based on stacked sparse denoising auto-encoder was proposed. The fault severity feature of rolling bearing is disturbed by the operating conditions and noise, the self-taught learning and feature extraction ability of the shallow network is low for fault feature. Therefore, the sparsity penalty term and denoising encoder were combined with auto-encoder, auto-encoder network was stacked and the classification layer was added to construct the stacked sparse denoising auto-encoder deep neural network,to achieve unsupervised feature extraction and intelligent diagnosis for rolling bearing. The ability of feature learning was improved by the sparsity penalty term and stacked auto-encoder, and the robustness of network was improved by the denoising encoder. The fault feature was automatically extracted and expressed to realize intelligent diagnosis, through training the layers individually with unsupervised and fining tuned with supervised.The feasibility and validity of the present method were verified byapplying the proposed method to diagnose the fault severity of rolling bearing under different operation conditions.

导出引用

陈仁祥1,2，杨星1，杨黎霞3，王家序2，徐向阳1，陈思杨1. 栈式稀疏加噪自编码深度神经网络的滚动轴承损伤程度诊断[J]. 振动与冲击, 2017, 36(21): 125-131

CHEN Ren-xiang1,2，YANG Xing1，YANG Li-Xia1，WANG Jia-xu2，XUXiang-yang1，CHEN Si-yang1. Fault severity diagnosis method forrolling bearingbased on stacked sparse denoising auto-encoder[J]. Journal of Vibration and Shock, 2017, 36(21): 125-131

参考文献

[1] 窦东阳,赵英凯.基于EMD 和Lempel-Ziv 指标的滚动轴承损伤程度识别研究[J].振动与冲击,2010,29(3):5-8.
DOU Dong-yang, ZHAO Ying-kai. Fault severity assessment for rolling element bearings based on EMD and Lempel-Ziv index[J]. Journal of Vibration and Shock,2010, 29(3):5-8.
[2] 从飞云,陈进,董广明.基于AR模型的Kolmogorov-Smirnov 检验性能退化及预测研究[J]. 振动与冲击,2012, 31(10):79-82.
CONG Fei-yun, CHEN Jin, DONG Guang-ming. Performance degradation assessment by Kolmogorov-Smirnov test and prognosis based on AR model [J]. Journal of Vibration and Shock,2012, 31(10):79 - 82.
[3] 张龙,黄文艺,熊国良.基于多尺度熵的滚动轴承故障程度评估[J].振动与冲击,2014,33(9):185-189.
ZHANG Long, HUANG Wen-yi, XIONG Guo-liang. Assessment of rolling element bearing fault severity using multi-scale entropy[J].Journal of Vibration and Shock,2014,33(9):85-189.
[4] Shuen-De,WuPo-Hung,WuChiu-Wen.Bearing Fault Diagnosis Based on Multiscale Permutation Entropy and Support Vector Machine[J].Entropy, 2012,14:1343-1356.
[5] 关贞珍,郑海起,叶明慧.基于振动图像纹理特征识别的轴承故障程度诊断方法研究[J].振动与冲击,2013,32(5):127-131.
GUAN Zhen-zhen, ZHENG Hai-qi,YE Ming-hui. Bearing fault severity assessment based on vibration image[J]. Journal of Vibration and Shock, 2013,32(5):127-131.
[6] Liu H,LiuC,Huang Y. Adaptive feature extraction usingsparse coding for machinery fault diagnosis[J]. Mechanical Systems and Signal Processing,2011,25(2):558-574.
[7] 朱会杰,王新晴,芮挺等.基于频域信号的稀疏编码在机械故障诊断中的应用[J].振动与冲击,2015,21(34):59-64.
ZHU Hui-jie, WANG Xin-qing, RUI Ting, etal. Application of sparse coding based on frequency domainsignals in machinery fault diagnosis.JOURNAL OF VIBRATION AND SHOCK,2015,21(34):59-64.
[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 Wenjun, SHAO Siyu,YANRuqiang. Induction Motor Fault Diagnosis Based on Deep Neural Network of Sparse Auto-encoder[J].JOURNAL OF MECHANICAL ENGINEERING,2016,52(9):65-71.
[9] 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,2016,21(51):49- 55.
LEI Yaguo,JIA Feng,ZHOUXin,etal.A Deep Learning- based Method for Machinery Health Monitoring with Big Data[J].JOURNAL OF MECHANICAL ENGINEERING, 2016,21(51):49-55.
[10] Tan CC, Eswaran C. Reconstruction and recognition of face and digit images using autoencoders[J]. Neural Computing and Applications, 2010,19(7):1069-1079.
[11] Vincent P,LarochelleH,LajoieI,et al. Stacked denoisingautoencoders: Learning useful representations in a deep network with a local denoising criterion[J]. Journal of Machine Learning Research,2010,11(6) : 3371-3408.
[12] BENGIO Y,LAMBLIN P,POPOVICI D.Greedylayerwise training of deep networks in Advances[C]//Proc of the 20thIn-ternational Conference on Neural Information Processing Systems. Cambridge: MIT Press,2007:153-159.
[13] VINCENT P,LAROCHELLE H,BENGIO Y, et al.Extracting and composing robust features with denoisingautoencoders[C] // Proceedings of the 25th International Conference on Machine Learning. July 5-9，2008，Helsinki，Finland：1096-1103.
[14] Feng Li, Jiaxu Wang, Baoping Tang, etal. Life grade recognition method based on supervised uncorrelated orthogonal locality preserving projection and K-nearest neighbor classifier[J].Neurocomputing,2014(138):271–282.