基于密集连接卷积网络和支持向量机的飞行器机械部件故障诊断

doi:10.13229/j.cnki.jdxbgxb20200702

摘要/Abstract

摘要：

针对旋转机械故障诊断浅层学习方法的高级特征提取问题和实际工程中可利用故障样本数量较少对诊断精度影响大的问题，提出了一种基于密集连接卷积网络（DenseNet）和支持向量机（SVM）的旋转机械故障诊断方法。首先，使用连续小波变换（CWT）将振动信号段转换为时频图像样本；然后，将试验样本输入DenseNet网络模型进行深层特征的提取；最后，将提取到的特征输入SVM模型进行训练，从而实现旋转机械的故障诊断。仿真结果表明：与其他先进模型相比，本文方法得到了更高的诊断准确率，证明了该方法的有效性和可行性。

关键词: 航空工程, 密集连接卷积网络, 支持向量机, 飞行器, 机械部件, 故障诊断

Abstract:

Aiming at the difficulty of deep feature extraction in the shallow learning methods of rotating machinery fault diagnosis and the problem of the small number of available fault samples in practical engineering， a rotating machinery fault diagnosis method based on densely connected convolutional network （DenseNet） and support vector machine （SVM） is proposed， which provides a new idea for the fault diagnosis of key mechanical components of aircraft. First， the continuous wavelet transform（CWT） is used to convert vibration signal segments into time-frequency image samples. Then， the experimental samples are input into the DenseNet model for feature extraction. Finally， the extracted features are used for the training of SVM， so as to realize the fault diagnosis of rotating machinery. The simulation results show that the proposed method has higher diagnostic performance than other advanced models， which proves the effectiveness and feasibility of the method.

Key words: aeronautical engineering, densely connected convolutional network, support vector machine, aircraft, mechanical components, fault diagnosis

中图分类号:

V267

院老虎,连冬杉,张亮,刘义. 基于密集连接卷积网络和支持向量机的飞行器机械部件故障诊断[J]. 吉林大学学报(工学版), 2021, 51(5): 1635-1641.

Lao-hu YUAN,Dong-shan LIAN,Liang ZHANG,Yi LIU. Fault diagnosis of key mechanical components of aircraft based on densenet and support vector machine[J]. Journal of Jilin University(Engineering and Technology Edition), 2021, 51(5): 1635-1641.

图/表 10

图1

图2

图3

图4

表1

图5

表2

表3

表4

表5

参考文献 25

1	The data comes from the PlaneCrashInfo.com database[EB/OL]. (2019-06-30). [2020-09-04].
2	Zhao Y P, Wang J J, Li X Y, et al. Extended least squares support vector machine with applications to fault diagnosis of aircraft engine[J]. ISA Transactions, 2020, 97: 189-201.
3	张翔, 姜爽, 赵岩, 等. 飞行器电源系统故障可观测性研究[J]. 电源技术, 2019, 43(3): 412-414.
	Zhang Xiang, Jiang Shuang, Zhao Yan, et al. Research on fault observability for aircraft power system[J]. Chinese Journal of Power Sources, 2019, 43(3): 412-414.
4	肖东, 江驹, 余朝军, 等. 基于EEMD分解和多分类支持向量机的飞行器舵面系统故障诊断[J]. 电光与控制, 2018, 25(8): 93-97.
	Xiao Dong, Jiang Ju, Yu Chao-jun, et al. Fault diagnosis of aircraft actuator system based on EEMD and multi-class SVM[J]. Electronics Optics & Control, 2018, 25(8): 93-97.
5	李国龙, 李彪, 蒋林, 等. 基于HHT和IPSO算法优化RBF神经网络的滚刀磨损状态识别方法[J]. 吉林大学学报: 工学版, 2020, 50(6): 1998-2009.
	Li Guo-long, Li Biao, Jiang Lin, et al. Hob wear recognition method based on HHT and IPSO-RBF neural network[J]. Journal of Jilin University(Engineering and Technology Edition), 2020, 50(6): 1998-2009.
6	Yan X A, Jia M P. A novel optimized SVM classification algorithm with multi-domain feature and its application to fault diagnosis of rolling bearing[J]. Neurocomputing, 2018, 313: 47-64.
7	Fujiyoshi H, Hirakawa T, Yamashita T. Deep learning-based image recognition for autonomous driving[J]. IATSS Research, 2019, 43(4): 244-252.
8	Khedkar S, Shinde S. Deep learning and ensemble approach for praise or complaint classification[J]. Procedia Computer Science, 2020, 167: 449-458.
9	Li Y, Zhang T, Sun S Y, et al. Accelerating flash calculation through deep learning methods[J]. Journal of Computational Physics, 2019, 394: 153-165.
10	姜洪开, 邵海东, 李兴球. 基于深度学习的飞行器智能故障诊断方法[J]. 机械工程学报, 2019, 55(7): 27-34.
	Jiang Hong-kai, Shao Hai-dong, Li Xing-qiu. Deep learning theory with application in intelligent fault diagnosis of aircraft[J]. Journal of Mechanical Engineering, 2019, 55(7): 27-34.
11	陈绵书, 于录录, 苏越, 等. 基于卷积神经网络的多标签图像分类[J]. 吉林大学学报: 工学版, 2020, 50(3): 1077-1084.
	Chen Mian-shu, Yu Lu-lu, Su Yue, et al. Multi-label images classification based on convolutional neural network[J]. Journal of Jilin University(Engineering and Technology Edition), 2020, 50(3): 1077-1084.
12	Yang Y T, Zheng H L, Li Y B, et al. A fault diagnosis scheme for rotating machinery using hierarchical symbolic analysis and convolutional neural network[J]. ISA Transactions, 2019, 91: 235-252.
13	Huang G, Liu Z, Laurens V D M, et al. Densely connected convolutional networks[C]∥IEEE Conference on Computer Vision and Pattern Recognition, Hawaii, 2017: 2261-2269.
14	Ahlawat S, Choudhary A. Hybrid CNN-SVM classifier for handwritten digit recognition[J]. Procedia Computer Science, 2020, 167: 2554-2560.
15	祝小彦, 王永杰, 张钰淇, 等. 基于自适应最优Morlet小波的滚动轴承故障诊断[J]. 振动、测试与诊断, 2018, 38(5): 1021-1029, 1085.
	Zhu Xiao-yan, Wang Yong-jie, Zhang Yu-qi, et al. Method of incipient fault diagnosis of bearing based on adaptive optimal Morlet wavelet[J]. Journal of Vibration, Measurement & Diagnosis, 2018, 38(5): 1021-1029, 1085.
16	He K M, Zhang X Y, Ren S Q, et al. Deep residual learning for image recognition[C]∥IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, 2016: 770-778.
17	范雪飞. 江苏千鹏诊断工程有限公司齿轮箱故障模拟试验数据[EB/OL]. (2012-06-07). [2020-09-04].
18	CWRU Dataset, Case Western Reserve University Bearing Data Center[EB/OL]. (2012-07-20). [2020-09-04].
19	张阳阳, 贾云献, 吴巍屹, 等. 概率神经网络在车辆齿轮箱典型故障诊断中的应用[J]. 汽车工程, 2020, 42(7): 972-977.
	Zhang Yang-yang, Jia Yun-xian, Wu Wei-yi, et al. Application of probabilistic neural network to typical fault diagnosis of vehicle gearbox[J]. Automotive Engineering, 2020, 42(7): 972-977.
20	龙霞飞, 杨苹, 郭红霞, 等. 基于KELM和多传感器信息融合的风电齿轮箱故障诊断[J]. 电力系统自动化, 2019, 43(17): 132-139.
	Long Xia-fei, Yang Ping, Guo Hong-xia, et al. Fault diagnosis of wind turbine gearbox based on KELM and multi-sensor information fusion[J]. Automation of Electric Power Systems, 2019, 43(17): 132-139.
21	李松柏, 康子剑, 陶洁. 基于信息融合及堆栈降噪自编码的齿轮故障诊断[J]. 振动与冲击, 2019, 38(5): 216-221.
	Li Song-bai, Kang Zi-jian, Tao Jie. Gear fault diagnosis based on information fusion and stacked de-noising auto-encoder[J]. Journal of Vibration and Shock, 2019, 38(5): 216-221.
22	Wang S H, Xiang J W, Zhong Y T, et al. Convolutional neural network-based hidden Markov models for rolling element bearing fault identification[J]. Knowledge-Based Systems, 2018, 144: 65-76.
23	Lu W N, Liang B, Cheng Y, et al. Deep model based domain adaptation for fault diagnosis[J]. IEEE Transactions on Industrial Electronics, 2017, 64(3): 2296-2305.
24	Li X, Zhang W, Ding Q. Cross-domain fault diagnosis of rolling element bearings using deep generative neural networks[J]. IEEE Transactions on Industrial Electronics, 2019, 66(7): 5525-5534.
25	Yuan L H, Lian D S, Kang X, et al. Rolling bearing fault diagnosis based on convolutional neural network and support vector machine[J]. IEEE Access, 2020, 8: 395-406.

相关文章 15

[1]	李伟,陈剑,陶善勇. 自适应耦合周期势系统随机共振信号增强方法[J]. 吉林大学学报(工学版), 2021, 51(3): 1091-1096.
[2]	欧阳丹彤,刘扬,刘杰. 故障响应指导下基于测试集的故障诊断方法[J]. 吉林大学学报(工学版), 2021, 51(3): 1017-1025.
[3]	王康,姚猛,李立犇,李建桥,邓湘金,邹猛,薛龙. 基于月面表取采样触月压痕的月壤力学状态分析[J]. 吉林大学学报(工学版), 2021, 51(3): 1146-1152.
[4]	潘凤文,弓栋梁,高莹,徐明伟,麻斌. 基于锂离子电池线性化模型的电流传感器故障诊断[J]. 吉林大学学报(工学版), 2021, 51(2): 435-441.
[5]	张根保,李浩,冉琰,李裘进. 一种用于轴承故障诊断的迁移学习模型[J]. 吉林大学学报(工学版), 2020, 50(5): 1617-1626.
[6]	李雄飞,王婧,张小利,范铁虎. 基于SVM和窗口梯度的多焦距图像融合方法[J]. 吉林大学学报(工学版), 2020, 50(1): 227-236.
[7]	谷远利, 张源, 芮小平, 陆文琦, 李萌, 王硕. 基于免疫算法优化LSSVM的短时交通流预测[J]. 吉林大学学报(工学版), 2019, 49(6): 1852-1857.
[8]	赵宏伟,李明昭,刘静,胡黄水,王丹,臧雪柏. 基于自然性和视觉特征通道的场景分类[J]. 吉林大学学报(工学版), 2019, 49(5): 1668-1675.
[9]	卢洋,王世刚,赵文婷,赵岩. 基于离散Shearlet类别可分性测度的人脸表情识别方法[J]. 吉林大学学报(工学版), 2019, 49(5): 1715-1725.
[10]	赵金钢,张明,占玉林,谢明志. 基于塑性应变能密度的钢筋混凝土墩柱损伤准则[J]. 吉林大学学报(工学版), 2019, 49(4): 1124-1133.
[11]	陈俊,张奇峰,张艾群,蔡笃思. 基于深渊鱼类识别的原位自主观测方法[J]. 吉林大学学报(工学版), 2019, 49(3): 953-962.
[12]	吴蔚楠,崔乃刚,郭继峰,赵杨杨. 多异构无人机任务规划的分布式一体化求解方法[J]. 吉林大学学报(工学版), 2018, 48(6): 1827-1837.
[13]	隗海林, 包翠竹, 李洪雪, 李明达. 基于最小二乘支持向量机的怠速时间预测[J]. 吉林大学学报(工学版), 2018, 48(5): 1360-1365.
[14]	王德军, 魏薇郦, 鲍亚新. 考虑侧风干扰的电子稳定控制系统执行器故障诊断[J]. 吉林大学学报(工学版), 2018, 48(5): 1548-1555.
[15]	耿庆田, 于繁华, 王宇婷, 高琦坤. 基于特征融合的车型检测新算法[J]. 吉林大学学报(工学版), 2018, 48(3): 929-935.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed

故障类型	实测转速/ （r·min^-1）	训练/测试样本个数	类别号
正常	880	30/8	1
大齿轮点蚀	880	30/8	2
大齿轮断齿	878	30/8	3
小齿轮磨损	881	30/8	4

模型	准确率/%	训练时间/s
本文	99.68±0.87	105.29
标准DenseNet	92.05±1.13	2028.00
标准SVM	84.38±12.5	26.07

类别	方法	诊断准确率/%
浅层学习	BPNN	89.06
	PNN	96.88
	GWO-KELM	95.58
深度学习	SDAE	95.17
深度学习	本文	99.68

故障类型	破坏直径/mm	样本个数	类别号
轴承正常	0.0000	76	1
内圈故障	0.1778	76	2
	0.3556	76	3
	0.5334	76	4
	0.7112	76	5
滚动体故障	0.1778	76	6
	0.3556	76	7
	0.5334	76	8
	0.7112	76	9
外圈故障	0.1778	76	10
	0.3556	76	11
	0.5334	76	12

方法	样本类型	准确率/%
CNN-HMM	原始数据	98.125
DAFD	频域信号	94.730
DGNN	频域信号	97.810
ResNet-SVM	时频图像	98.750
本文模型	时频图像	99.060