摘要轴承作为旋转机械中最易损耗的核心基础部件之一，是机械装备的重点监测对象。针对现有轴承智能故障诊断模型存在的对数据信息挖掘片面性及利用率低等问题，本文构建了一种基于双向长短期记忆（Bidirectional Long Short-term Memory）结构与多尺度卷积结构融合的深度学习网络模型。首先，为了增强模型的分类性能以及提高模型对实际工程环境的贴合度，数据集中各类故障数据的数据量为非等量；然后将数据集通过双向LSTM结构来获取具有对称性的数据特征，从而减少模型对前后故障信息记忆的紊乱、增强信息利用率，接着通过多尺度卷积结构对数据特征进行多角度理解与交流，防止特征提取片面化，同时还能增强模型的抗噪性能；最后通过全连接网络实现智能分类。将所提模型分别对深沟球轴承与圆柱滚子轴承故障数据进行处理分析，结果表明该智能模型具有较高的准确度与实用性。
关键词：双向长短期记忆；多尺度卷积；深度学习；轴承智能故障诊断
Abstract：Bearings, as one of the most wearable core basic components in rotating machinery, are the key monitoring objects of mechanical equipment. Aiming at the problems of one-sided data mining and low utilization of existing bearing intelligent fault diagnosis models, this paper constructs a deep learning network model based on the fusion of bidirectional long short-term memory structure and multi-scale convolution structure. First of all, in order to enhance the classification performance of the model and improve the fit of the model to the actual engineering environment, the data volume of various types of fault data in the data set is non-equal. Then, the data set is used to obtain symmetrical data features through a bidirectional LSTM structure. In this way, the model’s memory of the fault information before and after the fault is reduced, and the information utilization rate is enhanced. Then, the data features are understood and communicated from multiple angles through the multi-scale convolution structure to prevent the feature extraction from one-sidedness. Simultaneously, it can also enhance the anti-noise performance of the model. Finally, intelligent classification is realized through a fully connected network. The proposed model is used to process and analyze the fault data of deep groove ball bearings and cylindrical roller bearings. The results indicate that the intelligent model has high accuracy and practicability.
Key words: bidirectional long-short-term memory structure; multi-scale convolution structure; deep learning; intelligent fault diagnosis of bearing
欧阳励,何水龙,朱良玉,胡超凡,蒋占四. 一种基于双向长短期记忆结构与多尺度卷积结构融合的轴承智能故障诊断方法[J]. 振动与冲击, 2022, 41(19): 179-187.
OUYANG Li, HE Shuilong, ZHU Liangyu, HU Chaofan, JIANG Zhansi. An intelligent bearing fault diagnosis method based on fusion of bidirectional LSTM structure and MSC structure. JOURNAL OF VIBRATION AND SHOCK, 2022, 41(19): 179-187.
[1] 胡超凡,王衍学.基于张量分解的滚动轴承复合故障多通道信号降噪方法研究[J].机械工程学报,2019,55(12):50-57.
Hu Chaofan, WANG Yanxue. Research on Multi-channel Signal Denoising Method for Multiple Faults Diagnosis of Rolling Element Bearings Based on Tensor Factorization [J]. Journal of Mechanical Engineering, 2019, 55(12): 50-57.
[2] 王建国,李健,万旭东.基于奇异值分解和局域均值分解的滚动轴承故障特征提取方法[J].机械工程学报, 2015, 51(03): 104-110.
WANG Jianguo, LI Jian, WAN Xudong. Fault Feature Extraction Method of Rolling Bearings Based on Singular Value Decomposition and Local Mean Decomposition [J]. Journal of Mechanical Engineering, 2015, 51(03):104-110.
[3] WANG Yanxue，HE Zhengjia，ZI Yanyang. Enhancement of signal denoising and multiple fault signatures detecting in rotating machinery using dual-tree complex wavelet transform [J]. Mechanical Systems and Signal Processing，2010，24(1)：119-137.
[4] WANG Yanxue，LIANG Ming. An adaptive SK technique and its application for fault detection of rolling element bearings [J]. Mechanical Systems and Signal Processing，2011，25 (5)：1750-1764.
[5] LECUN Y, BENGIO Y, HINTON G. Deep learning [J]. Nature, 2015, 521(7553): 436-440.
[6] Hu Chaofan, Wang Yanxue, Bai T. A Tensor-Based Approach for Identification of Multi-Channel Bearing Compound Faults [J]. IEEE Access, 2019, 7: 38213-38223.
[7] Hu Chaofan, Wang Yanxue. Multidimensional denoising of rotating machine based on tensor factorization [J]. Mechanical Systems and Signal Processing, 2019, 122: 273-289.
[8] Hu Chaofan, Wang Yanxue, Gu jiawei. Cross-domain intelligent fault diagnosis of rolling element bearings based on tensor-aligned invariant subspace learning and two-dimensional convolutional neural networks [J]. Knowledge Based Systems, 2020 (209).
[9] Hu Chaofan, Wang Yanxue, He shuilong. A classification method to detect faults in a rotating machinery based on kernelled support tensor machine and multilinear principal component analysis [J]. Applied Intelligence, 2020,1-13.
[10] 薛妍,沈宁,窦东阳.基于一维卷积神经网络的滚动轴承故障程度诊断[J].轴承,2021(04):48-54.
XUE Yan, SHEN Ning, DOU Dongyang. Fault Degree Diagnosis of Rolling Bearings Based on One- Dimensional Convolutional Neural Network [J]. Bearing, 2021(04):48-54.
[11] 曲建岭, 余路, 袁涛等. 基于一维卷积神经网络的滚动轴承自适应故障诊断算法[J]. 仪器仪表学报, 2018, 39(07): 134-143.
Qu Jianling, Yu Lu, Yuan Tao, et al. Adaptive fault diagnosis algorithm for rolling bearings based on one-dimensional convolutional neural network [J]. Chinese Journal of Scientific Instrument, 2018, 39(07): 134-143.
[12] 庞俊,刘鑫,段敏霞等.基于改进卷积神经网络轴承故障诊断[J].组合机床与自动化加工技术,2021(03):66-69.
PANG Jun，LIU Xin，DUAN Min xia, et al. Fault Diagnosis Based on Improved Convolutional Neural Network [J]. Modular Machine Tool ＆ Automatic Manufacturing Technique, 2021(03):66-69.
[13] 李自纳,唐银敏,吴延艳.基于长短时记忆循环网络的塑料编织机故障诊断研究[J].塑料科技,2020,48(10):86-89.
LI Zina, TANG Yinmin, WU Yanyan. Research on Rault Diagnosis of Plastic Braiding Machine Based on Long-Short Memory Convolutional Neural Network [J]. Plastics Science and Technology, 2020, 48(10):86-89.
[14] 池永为,杨世锡,焦卫东.基于LSTM-RNN的滚动轴承故障多标签分类方法[J].振动.测试与诊断,2020,40(03):563-571+629.
CHI Yongwei, YANG Shixi, JIAO Weidong. A Multi-label Fault Classification Method for Rolling Bearing Based on LSTM-RNN [J]. Journal of Vibration, Measurement & Diagnosis, 2020, 40(03):563-571+629.
[15] 张龙,徐天鹏,王朝兵,易剑昱,甄灿壮.基于卷积门控循环网络的齿轮箱故障诊断[J/OL].吉林大学学报(工学版): 1-10[2021-04-22].https://doi.org/10.13229/j.cnki.jdxbgxb20200930.
ZhANG Long, XU Tianpeng, WANG Chaobing, et al. Gearbox fault diagnosis based on convolutional gated recurrent network [J/OL]. Journal of Jilin University (Engineering and Technology Edition):1-10 [2021-04-22]. https://doi.org/10. 13229/ j. cnki. jdxbgxb20200930.
[16] 张立鹏,毕凤荣,程建刚,沈鹏飞.基于注意力BiGRU的机械故障诊断方法研究[J].振动与冲击,2021,40(05):113-118.
ZHANG Lipeng, BI Fengrong, CHENG Jiangang, et al. Mechanical fault diagnosis method based on attention BiGRU [J]. Journal of Vibration and Shock, 2021, 40(05):113-118.
[17] 郑直,张华钦,潘月.基于改进鲸鱼算法优化LSTM的滚动轴承故障诊断[J].振动与冲击,2021,40(07):274-280.
ZHENG Zhi, ZHANG Huaqin, PAN Yue. Rolling bearing fault diagnosis based on IWOA-LSTM [J]. Journal of Vibration and Shock, 2021,40(07):274-280.
[18] 鲁其东,程丽平,杨瑞,钟麦英. 基于循环卷积神经网络的轴承故障诊断[A]. 中国自动化学会过程控制专业委员会.第30届中国过程控制会议（CPCC 2019）摘要集[C].中国自动化学会过程控制专业委员会:中国自动化学会过程控制专业委员会,2019:1.
LU Qidong, CHENG Liping, YANG Rui, et al. Bearing Fault Diagnosis based on R-CNN [A]. The Process Control Professional Committee of the Chinese Society of Automation. The 30th China Process Control Conference (CPCC 2019) Abstract Collection [C]. The Process Control Professional Committee of the Chinese Society of Automation: The Process Control Professional Committee of the Chinese Society of Automation, 2019:1.
[19] Hochreiter S, Schmidhuber J. Long short-term memory [J]. Neural computation, 1997, 9(8): 1735-1780.
[20] Vaswani A, Shazeer N, Parmar N, et al. Attention is all you need [J]. arXiv preprint arXiv, 2017, 1706.03762.
