基于一维CNN参数优化的压缩振动信号故障诊断

doi:10.3969/j.issn.1001-506X.2020.09.05

摘要/Abstract

摘要：

针对一维卷积神经网络(convolutional neural network, CNN)参数多的特点,提出一种正交试验和粒子群优化算法相结合的参数优化方法,并将其应用于压缩振动信号故障诊断。压缩感知理论突破了奈奎斯特采样定理的限制,为大量振动信号的采集与传输提供一种有效途径。首先利用CNN“端-端”特性,建立了基于压缩信号的CNN故障诊断模型。利用正交试验进行参数范围的粗略评价,选择出最优方案。对最优方案中每个参数,利用多目标粒子群优化算法进行细化,得出精确的参数最优取值。选择齿轮箱实测信号和西储大学轴承信号作为研究对象。实验结果表明，经过优化后非劣粒子的输出特征分类明显, CNN诊断率有明显提高，也证明了对压缩信号直接进行故障诊断的可行性。

关键词: 卷积神经网络, 压缩感知, 多目标粒子群, 正交试验, 故障诊断

Abstract:

In view of the characteristics of multi-parameter for the 1-dimensional convolutional neural network (CNN), a parameter optimization method combining orthogonal test and particle swarm optimization algorithm is proposed and applied to the fault diagnosis of compressed vibration signals. Compressive sensing theory breaks through the limitation of Nyquist sampling theorem and provides an effective way for the collection and transmission of a large number of vibration signals. Firstly, a CNN fault diagnosis model based on compressed signal is established by using the "end-to-end" feature of CNN. The orthogonal test is used to make a rough evaluation of the parameter range, and the best scheme is selected. For each parameter in the scheme, the multi-objective particle swarm optimization algorithm is used to refine it, and the optimal value of each parameter is obtained. The measured signal of gear box and bearing signal from Case Western Reserve University are selected as the research objects. The experimental results show that after optimization, the output characteristic classification of non-inferior particles is obvious, and the CNN diagnosis rate is significantly improved. The results also prove the feasibility of direct fault diagnosis for compressed signals.

Key words: convolutional neural network (CNN), compressive sensing, multi-objective particle swarm, orthogonal experiment, fault diagnosis

中图分类号:

马云飞, 贾希胜, 白华军, 郭驰名, 王双川. 基于一维CNN参数优化的压缩振动信号故障诊断[J]. 系统工程与电子技术, 2020, 42(9): 1911-1919.

Yunfei MA, Xisheng JIA, Huajun BAI, Chiming GUO, Shuangchuan WANG. Fault diagnosis of compressed vibration signal based on 1-dimensional CNN with optimized parameters[J]. Systems Engineering and Electronics, 2020, 42(9): 1911-1919.

图/表 16

图1

图2

图3

图4

图5

表1

表2

表3

图6

图7

表4

图8

表5

图9

图10

表6

参考文献 30

1	HINTON G E , SALAKHUTDINOV R R . Reducing the dimensionality of data with neural network[J]. Science, 2006, 313 (5786): 504- 507.
2	周奇才, 刘星辰, 赵炯, 等. 旋转机械一维深度卷积神经网络故障诊断研究[J]. 振动与冲击, 2018, 37 (23): 31- 37.
	ZHOU Q C , LIU X C , ZHAO J , et al. Fault diagnosis for rotating machinery based on 1D depth convolutional neural network[J]. Journal of Vibration and Shock, 2018, 37 (23): 31- 37.
3	赵春华, 胡恒星, 陈保家, 等. 基于深度学习特征提取和WOA-SVM状态识别的轴承故障诊断[J]. 振动与冲击, 2019, 38 (10): 31- 37, 48.
	ZHAO C H , HU H X , CHEN B J , et al. Bearing fault diagnosis based on deep learning feature extraction and WOA-SVM state recognition[J]. Journal of Vibration and Shock, 2019, 38 (10): 31- 37, 48.
4	SHAO H D , JIANG H K , ZHAO H W , et al. A novel deep autoencoder feature learning method for rotating machinery fault diagnosis[J]. Mechanical Systems & Signal Processing, 2017, 95, 187- 204.
5	曲建岭, 余路, 袁涛, 等. 基于卷积神经网络的层级化智能故障诊断算法[J]. 控制与决策, 2019, 34 (12): 2619- 2626.
	QU J L , YU L , YUAN T , et al. A hierarchical intelligent fault diagnosis algorithm based on convolutional neural network[J]. Control and Decision, 2019, 34 (12): 2619- 2626.
6	FENG J , LEI Y G , LIN J , et al. Deep neural networks: a promising tool for fault characteristic mining and intelligent diagnosis of rotating machinery with massive data[J]. Mechanical Systems & Signal Processing, 2016, 72/73, 303- 315.
7	LU C , WANG Z Y , QIN W L , et al. Fault diagnosis of rotary machinery components using a stacked denoising autoencoder-based health state identification[J]. Signal Processing, 2017, 130, 377- 388.
8	GAN M , WANG C , ZHU C A . Construction of hierarchical diagnosis network based on deep learning and its application in the fault pattern recognition of rolling element bearings[J]. Mechanical Systems & Signal Processing, 2016, 72/73, 92- 104.
9	任浩, 屈剑锋, 柴毅, 等. 深度学习在故障诊断领域中的研究现状与挑战[J]. 控制与决策, 2017, 32 (8): 345- 358.
	REN H , QU J F , CHAI Y , et al. Deep learning for fault diagnosis: The state of the art and challenge[J]. Control and Decision, 2017, 32 (8): 345- 358.
10	KRIZHEVSKY A , SUTSKEVER I , HINTON G E . ImageNet classification with deep convolutional neural networks[J]. Advances in neural information processing systems, 2012, 25 (2): 1097- 1105.
11	SZEGEDY C, LIU W, JIA Y, et al. Going deeper with convolutions[C]//Proc.of the IEEE Conference on Computer Vision and Pattern Recognition, 2015: 1-9.
12	HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition[C]//Proc.of the Computer Vision and Pattern Recognition, 2016: 770-778.
13	CHEN Z Q , LI C , SANCHEZ R V . Gearbox fault identification and classification with convolutional neural networks[J]. Shock and Vibration, 2015, (2): 390134.
14	ZHANG W, PENG G L, LI C H. Rolling element bearings fault intelligent diagnosis based on convolutional neural networks using raw sensing signal[C]//Proc.of the 12th International Conference on Intelligent Information Hiding and Multimedia Signal Processing, 2017: 77-84.
15	INCE T , KIRANYAZ S , EREN L , et al. Real-time motor fault detection by 1D convolutional neural networks[J]. IEEE Trans.on Industrial Electronics, 2016, 63 (11): 7067- 7075. doi: 10.1109/TIE.2016.2582729
16	吴春志, 江鹏程, 冯辅周, 等. 基于一维卷积神经网络的齿轮箱故障诊断[J]. 振动与冲击, 2018, 37 (22): 51- 56.
	WU C Z , JIANG P C , FENG F Z , et al. Faults diagnosis method for gearboxes based on a 1-D convolutional neural network[J]. Journal of Vibration and Shock, 2018, 37 (22): 51- 56.
17	朱会杰, 王新晴, 芮挺, 等. 基于平移不变CNN的机械故障诊断研究[J]. 振动与冲击, 2019, 38 (5): 45- 52.
	ZHU H J , WANG X Q , RUI T , et al. Machinery fault diagnosis based on shift invariant CNN[J]. Journal of Vibration and Shock, 2019, 38 (5): 45- 52.
18	CANDES E J. Compressive sampling[C]//Proc.of International Congress of Mathematics, 2006, 3: 1433-1452.
19	CANDÈS E J , TAO T . Decoding by linear programming[J]. IEEE Trans.on Information Theory, 2005, 51 (12): 4203- 4215.
20	DONOHO D L . Compressed sensing[J]. IEEE Trans.on Information Theory, 2006, 52 (4): 1289- 1306. doi: 10.1109/TIT.2006.871582
21	TROPP J A , GILBERT A C . Signal recovery from random measurements via orthogonal matching pursuit[J]. IEEE Trans.on Information Theory, 2007, 53 (12): 4655- 4666.
22	ZHANG X P , HU N Q , HU L , et al. A bearing fault diagnosis method based on the low dimensional compressed vibration signal[J]. Advances in Mechanical Engineering, 2015, 7 (7): 1- 12.
23	WANG Y X , XIANG J W , MO Q Y , et al. Compressed sparse time-frequency feature representation via compressive sensing and its applications in fault diagnosis[J]. Measurement, 2015, 68, 70- 81.
24	TANG G , HOU W , WANG H Q , et al. Compressive sensing of roller bearing faults via harmonic detection from under-sampled vibration signals[J]. sensors, 2015, 15 (10): 25648- 25662.
25	温江涛, 闫常弘, 孙洁娣, 等. 基于压缩采样与深度学习的轴承故障诊断方法[J]. 仪器仪表学报, 2018, 39 (1): 171- 179.
	WEN J T , YAN C H , SUN J D , et al. Bearing fault diagnosis method based on compressed acquisition and deep learning[J]. Chinese Journal of Scientific Instrument, 2018, 39 (1): 171- 179.
26	MA Y F , JIA X S , BAI H J , et al. A new fault diagnosis method based on convolutional neural network and compressive sensing[J]. Journal of Mechanical Science and Technology, 2019, 33, 5177- 5188.
27	LECUN Y , BOTTOU L , BENGIO Y , et al. Gradient-based learning applied to document recognition[J]. Proceedings of the IEEE, 1998, 86 (11): 2278- 2324.
28	方开泰, 刘民千, 周永道. 试验设计与建模[M]. 北京: 高等教育出版社, 2011: 81- 114.
	FANG K T , LIU M Q , ZHOU Y D . Design and modeling of experiments[M]. Beijing: Higher Education Press, 2011: 81- 114.
29	张新峰, 焦月, 李欢欢, 等. 基于粒子群算法的Universum SVM参数选择[J]. 北京工业大学学报, 2013, 39 (6): 840- 845.
	ZHANG X F , JIAO Y , LI H H , et al. Model paremeter selection of the universum SVM based on particle swarm optimazation[J]. Journal of Beijing University of Technology, 2013, 39 (6): 840- 845.
30	COELLO C A C , PULIDO T G , LECHUGA M S . Handling multiple objectives with particle swarm optimization[J]. Evolutionary Computation, 2004, 8 (3): 256- 279.

状态	负载/Nm	转速/rpm	训练样本	测试样本
正常状态	0.4	800	100	20
正常状态	0.8	1 200	100	20
行星轮故障	0.4	800	100	20
行星轮故障	0.8	1 200	100	20
齿圈故障	0.4	800	100	20
齿圈故障	0.8	1 200	100	20
太阳轮故障	0.4	800	100	20
太阳轮故障	0.8	1 200	100	20

序号	卷积核个数A	卷积核大小B	批尺寸C	池化层大小D	准确率/%	时间/s
1	5	50	5	2	76.9	74.99
2	5	100	10	5	93.1	43.81
3	5	200	20	10	82.5	35.41
4	10	50	10	10	87.5	82.92
5	10	100	20	2	63.1	223.78
6	10	200	5	5	95.0	194.43
7	15	50	20	5	60.0	262.63
8	15	100	5	10	92.5	256.65
9	15	200	10	2	60.6	628.43

序号	卷积核个数A	卷积核大小B	批尺寸C	池化层大小D
T₁	252.5%(154.21)	224.4%(420.54)	264.4%(526.07)	200.6%(927.2)
T₂	245.6%(501.13)	248.7%(524.24)	241.2%(755.16)	248.1%(500.87)
T₃	213.1%(1 147.71)	238.1(858.27)	205.6%(521.82)	262.5%(374.98)
m₁	84.2%(51.40)	74.8%(140.18)	88.1%(175.36)	66.9%(309.07)
m₂	81.9%(167.04)	82.9%(174.75)	80.4%(251.72)	82.7%(166.96)
m₃	71.0%(382.57)	79.4%(286.09)	68.5%(173.94)	87.5%(124.99)
R	13.2%(331.17)	8.1%(145.91)	19.6%(77.78)	20.6%(184.08)

优化目标	文献[26]	正交试验优化	正交试验+PSO
诊断成功率/%	97.5	98.1	99.4
运行时间/s	189	184.27	192.8

数据样本	故障深度/mm	电机功率/HP	电机转速/rpm
1(N)	-	0/1/2/3	1797/1772/1750/1730
2(IR)	7	0/1/2/3	1797/1772/1750/1730
3(B)	7	0/1/2/3	1797/1772/1750/1730
4(OR)	7	0/3/1/2	1797/1730/1772/1750
5(IR)	14	0/1/2/3	1797/1772/1750/1730
6(B)	14	0/1/2/3	1797/1772/1750/1730
7(OR)	14	0/1/2/3	1797/1772/1750/1730
8(IR)	21	0/1/2/3	1797/1772/1750/1730
9(B)	21	0/1/2/3	1797/1772/1750/1730
10(OR)	21	1/3/0/2	1772/1730/1797/1750