针对当前基于深度学习的航空发动机滚动轴承故障诊断技术诊断任务单一的问题，提出一种基于多任务残差网络的滚动轴承故障诊断方法，该方法采用残差网络为深层特征提取与共享主框架，建立能够同时进行故障诊断的多任务模型。首先，在数据预处理中，将滚动轴承的振动加速度时域信号转换为频谱图，并直接作为网络的输入；然后，应用标签平滑技术对故障类别标签做了平滑处理以提高网络的测试精度；最后，利用两组实际的滚动轴承故障数据集对所建立的多任务模型进行试验验证，将诊断任务划分为：故障状态识别（正常和异常）、故障部位识别（内圈、外圈和滚动体故障）、以及故障程度识别（损伤尺寸大小预测）。结果表明，所搭建的多任务模型在故障状态识别和部位诊断中的准确率达到97%以上。同时，在故障识别中，损伤大小预测达到了满意的精度，充分表明该方法具有很强的故障多任务诊断能力。
关键词：深度学习；残差网络；多任务；滚动轴承；故障诊断；损伤大小
Abstract
The current technology in the diagnosis of rolling bearing fault diagnosis based on deep learning task to a single problem, this paper proposes a multitasking residual network based fault diagnosis method of rolling bearing, the method adopts the deep residual network for feature extraction and share the main frame, to establish model of fault diagnosis of many tasks at the same time, first of all, in data preprocessing,The time domain signal of vibration acceleration of rolling bearing is converted into a spectrum graph and directly used as the input of the network.Then the fault category labels are smoothed by label smoothing technique to improve the testing accuracy of the network.Finally, two sets of actual rolling bearing fault data sets were used to verify the established multi-task model, and the diagnosis tasks were divided into: fault state identification (normal and abnormal), fault position identification (inner ring, outer ring and rolling body faults), and fault degree identification (damage size prediction).The results show that the accuracy of the proposed multi-task model in fault state identification and location diagnosis reaches more than 97%. Meanwhile, the damage size prediction achieves satisfactory accuracy in fault identification, which fully shows that the proposed method has strong multi-task fault diagnosis capability.
Key words：deep learning;residual network; multitasking;rolling bearing;fault diagnosis;damage size
关键词
深度学习 /
残差网络 /
多任务 /
滚动轴承 /
故障诊断 /
损伤大小
{{custom_keyword}} /
Key words
deep learning;residual network /
multitasking;rolling bearing;fault diagnosis;damage size
{{custom_keyword}} /
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1]. 张向阳,陈果,郝腾飞.等.基于机匣信号的滚动轴承故障卷积神经网络诊断方法[J].航空动力学报,2019,34(12):2729-2737.
ZHANG Xiangyang, CHEN Guo, HAO Tengfei, et al. Convolutional neural network diagnosis method of rolling bearing fault based on casing signal[J]. Journal of Aerospace Power, 2019, 34(12): 2729-2737.
[2]. LEI Y G,YANG B,JIANG X W, et al. Applications of machine learning to machine fault diagnosis: a review and roadmap[J]. Mechanical Systems and Signal Processing,2020,138: 106587.
[3]. 陈果.滚动轴承早期故障的特征提取与智能诊断[J].航空学报,2009,30(2):362-367.
CHEN Guo. Feature extraction and intelligent diagnosis for ball bearing early faults[J]. Acta Aeronauticaet Astronautica Sinica, 2009, 30(2): 362-367.
[4]. 田科位，董绍江，姜保军，等. 基于改进深度残差网络的轴承故障诊断方法[J]. 振动与冲击, 2021, 40(20): 247-254.
TIAN Kewei, DONG Shaojiang, JIANG Baojun, et al. A bearing fault diagnosis method based on an improved depth residual network[J]. Journal of Vibration and Shock, 2021, 40(20): 247-254.
[5]. ZHOU F N , Yang S , HAMIDO F, et al. Deep learning fault diagnosis method based on global optimization GAN for unbalanced data[J]. Knowledge-Based Systems, 2020, 187:104837.
[6]. 孟宗, 关阳, 潘作舟,等. 基于二次数据增强和深度卷积的滚动轴承故障诊断研究[J]. 机械工程学报, 2021, 57(23): 106-115.
MENG Zong, GUAN Yang, PAN Zuozhou, et al. Fault diagnosis of rolling bearing based on secondary data enhancement and deep convolutional network[J]. Journal of Mechanical Engineering, 2021, 57(23): 106-115.
[7]. 王琦,邓林峰,赵荣珍.基于改进一维卷积神经网络的滚动轴承故障识别[J].振动与冲击,2022,41(3):216-223.
WANG Qi, DENG Linfeng, ZHAO Rongzhen. Fault recognition of rolling bearing based on improved 1D convolutional neural network[J].Journal of Vibration and Shock, 2022, 41(3): 216-223.
[8]. 刘飞,陈仁文,邢凯玲,等.基于迁移学习与深度残差网络的滚动轴承快速故障诊断算法[J].振动与冲击,2022,41(3):154-164.
LIU Fei, CHEN Renwen, XING Kailing, et al. Fast fault diagnosis algorithm for rolling bearing based on transfer learning and deep residual network[J].Journal of Vibration and Shock, 2022, 41(3): 154-164.
[9]. LEI J H, LIU C,JIANG D X. Fault diagnosis of wind turbine based on long short-term memory networks[J]. Renewable Energy,2019,133:422-432.
[10]. SHAO H D,JIANG H K, ZHANG H Z , et al. Rolling bearing fault feature learning using improved convolutional deep belief network with compressed sensing[J]. Mechanical Systems and Signal Processing, 2018, 100:743-765.
[11]. ??NGUYEN, NGOC H , CHEOL-HONG, et al. effective prediction of bearing fault degradation under different crack sizes using a deep neural network.
[12]. 王震,黄如意,李霁蒲,等.一种用于故障分类与预测的多任务特征共享神经网络[J].仪器仪表学报,2019,40(7):169-177.
WANG Zhen, HUANG Ruyi,LI Jipu, et al.Multi-task feature sharing neural network used for fault diagnosis and prognosis[J].Chinese Journal of Scientific Instrument,2019,40(7):169-177.
[13]. HE K M , ZHANG X Y , REN S Q, et al. Deep residual learning for image recognition[C]// 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Vegas:IEEE, 2016.
[14]. HE K M , ZHANG X Y , REN S Q,et al. Identity mappings in deep residual networks[C]// European Conference on Computer Vision. Cham: Springer, 2016.
[15]. LECUN Y, BOSER B, DENKER J S,et al. Backpropagation applied to handwritten zip code recognition[J]. Neural Computation,1989,1(4)：541-551.
[16]. ？？MÜLLER R, KORNBLITH S, HINTON G . When does label smoothing help?. 32, 4696-4705.
[17]. ？？Bearing data center seeded fault test data. The Case Western Reserve University Bearing Data CenterWebsite.
{{custom_fnGroup.title_cn}}
脚注
{{custom_fn.content}}
PDF(1775 KB)
