基于多尺度AlexNet网络的健康因子构建方法

doi:10.3969/j.issn.1001-506X.2020.01.33

摘要/Abstract

摘要：

针对目前健康因子构建方法存在的单调性和趋势性不够理想的问题，提出一种基于多尺度AlexNet网络的轴承健康因子构建方法。该方法首先利用连续小波分析将原始振动加速度信号转换为时频图，将时频图作为输入对多尺度AlexNet网络进行训练；然后利用训练好的网络在线构建测试轴承健康因子；最后根据健康因子评估准则评估初步构建的健康因子，利用评估结果调整网络参数，实现迭代优化，进一步提高健康因子的单调性和趋势性。实验对比分析结果表明：该方法显著提高了健康因子的单调性与趋势性，不需要进行特征提取、特征选择、特征融合等步骤，具有较高的构建效率和泛化性。

关键词: 轴承健康因子, 深度学习, 卷积神经网络, 评估准则

Abstract:

The monotonicity and trendability of the health indicator constructed by the traditional method is not very ideal. To solve these problems, a bearing health indicator construction method based on multi-scale AlexNet network is proposed. First of all, the original vibrational acceleration signal is transformed to time-frequency map, which is considered as the input of the multi-scale AlexNet network, by continuous wavelet transformation. Then, the health indicator of test bearing is constructed online by the trained multi-scale AlexNet network. Finally, the constructed health indicator is evaluated by metrics, the evaluation results are then used to adjust the network parameters to realize the iterative optimization. The experiment results show that the health indicator constructed by the proposed health indicator construction method has better trendability and monotonicity. In addition, this method does not rely on feature extraction, selection and fusion, which enhances the construction efficiency and generalization.

Key words: bearing health indicator, deep learning, convolutional neural network, evaluation metrics

中图分类号:

TN911.72

张钢, 田福庆, 梁伟阁, 佘博. 基于多尺度AlexNet网络的健康因子构建方法[J]. 系统工程与电子技术, 2020, 42(1): 245-252.

Gang ZHANG, Fuqing TIAN, Weige LIANG, Bo SHE. Construction method of bearing health indicator based on multi-scale AlexNet network[J]. Systems Engineering and Electronics, 2020, 42(1): 245-252.

图/表 12

图1

图2

图3

表1

表2

图4

图5

表3

图6

图7

图8

表4

参考文献 22

1	CERRADA M , SANCHEZ R V , LI C , et al. A review on data-driven fault severity assessment in rolling bearings[J]. Mechanical Systems and Signal Processing, 2018, 99, 169- 196. doi: 10.1016/j.ymssp.2017.06.012
2	LEI Y G , LI N P , GUO L , et al. Machinery health prognostics:a systematic review from data acquisition to RUL prediction[J]. Mechanical Systems and Signal Processing, 2018, 104, 799- 834. doi: 10.1016/j.ymssp.2017.11.016
3	袁广超, 鲍劲松, 郑小虎, 等. 基于CNC实时监测数据驱动方法的钛合金高速铣削刀具寿命预测[J]. 中国机械工程, 2018, 29 (4): 457- 462. doi: 10.3969/j.issn.1004-132X.2018.04.013
	YUAN G C , BAO J S , ZHENG X H , et al. Tool life prediction in titanium high speed milling processes based on CNC real time monitoring data driven[J]. China Mechanical Engineering, 2018, 29 (4): 457- 462. doi: 10.3969/j.issn.1004-132X.2018.04.013
4	WANG D , TSUI K L , MIAO Q . Prognostics and health management:a review of vibration based bearing and gear health indicators[J]. IEEE Access, 2018, 6, 665- 676.
5	CHAO H , BYENG D Y , WANG P F , et al. Ensemble of data-driven prognostic algorithms for robust prediction of remaining useful life[J]. Reliability Engineering and System Safety, 2012, 103, 120- 135. doi: 10.1016/j.ress.2012.03.008
6	HONG S , ZHOU Z , ZIO E , et al. An adaptive method for health trend prediction of rotating bearings[J]. Digital Signal Processing, 2014, 35, 117- 123. doi: 10.1016/j.dsp.2014.08.006
7	JAVED K , GOURIVEAU R , ZERHOUNI N . Enabling health monitoring approach based on vibration data for accurate prognostics[J]. IEEE Trans.on Industrial Electronics, 2015, 62 (1): 648- 656.
8	WIDODO A , YANG B S . Application of relevance vector machine and survival probability to machine degradation assessment[J]. Expert Systems with Applications, 2011, 38 (3): 2592- 2599.
9	SOUALHI A , MEDJAHER K , ZERHOUNI N . Bearing health monitoring based on Hilbert-Huang transform, support vector machine and regression[J]. IEEE Trans.on Instrumentation and Measurement, 2015, 64 (1): 52- 62. doi: 10.1109/TIM.2014.2330494
10	吴春志, 贾继德, 贾翔宇. 基于多尺度降维的柴油机信号信息熵增强方法[J]. 振动与冲击, 2018, 37 (3): 180- 185.
	WU C Z , JIA J D , JIA X Y . A method to enhance information entropy of diesel engine signals based on multi-scale dimention reduction[J]. Journal of Vibration and Shock, 2018, 37 (3): 180- 185.
11	LEI Y G , LI N P , LIN J . A model-based method for remaining useful life prediction of machinery[J]. IEEE Trans.on Reliability, 2016, 65 (3): 1314- 1326. doi: 10.1109/TR.2016.2570568
12	孙健, 李洪儒. 一种基于复合谱与关联熵融合的特征提取方法[J]. 机械工程学报, 2017, 53 (24): 96- 103.
	SUN J , LI H R . Method for feature extraction based on composite spectrum and relative entropy fusion[J]. Journal of Mechanical Engineering, 2017, 53 (24): 96- 103.
13	BENKEDJOUH T , MEDJAHER K , ZERHOUNI N , et al. Health assessment and life prediction of cutting tools based on support vector regression[J]. Journal of Intelligent Manufacturing, 2015, 26, 213- 223. doi: 10.1007/s10845-013-0774-6
14	LIAO L X , WEN J , PAVEL R . Enhanced restricted boltzmann machine with prognosability regularization for prognostics and health assessment[J]. IEEE Trans.on Industrial Eletronics, 2016, 63 (11): 7076- 7083. doi: 10.1109/TIE.2016.2586442
15	WANG Y , PENG Y , ZI Y , et al. A two-stage data-driven-based prognostic approach for bearing degradation problem[J]. IEEE Trans.on Industrial Informatics, 2016, 12, 924- 932. doi: 10.1109/TII.2016.2535368
16	赵光权, 刘小勇, 姜泽东, 等. 基于深度学习的轴承健康因子无监督构建方法[J]. 仪器仪表学报, 2018, 39 (6): 82- 88.
	ZHAO G Q , LIU X Y , JIANG Z D , et al. Unsupervised health indicator of bearing based on deep learning[J]. Chinese Journal of Scientific Instrument, 2018, 39 (6): 82- 88.
17	GUO L , LEI Y G , LI N P , et al. Machinery health indicator construction based on convolutional neural networks considering trnd blur[J]. Neurocomputing, 2018, 292, 142- 150. doi: 10.1016/j.neucom.2018.02.083
18	MARCO R , PIERO B , ENRICO Z , et al. Ensemble of optimized echo state networks for remaining useful life prediction[J]. Neurocomputing, 2018, 281, 121- 138. doi: 10.1016/j.neucom.2017.11.062
19	KRIZHEVSKY A , SUTSKEVER I , HINTON G . Imagenet classification with deep convolutional neural networks[J]. Advances in Neural Information Processing Systems, 2012, 25 (2): 1097- 1105.
20	BERGSTRA J , BENGIO Y . Random search for hyper-parameter optimization[J]. Journal of Machine Learning Research, 2012, 13 (1): 281- 305.
21	NECTOUX P, GOURIVEAU R, MEDJAHER K, et al. PRONOSTIA: an experimental platform for bearings accelerated degradation tests[C]//Proc.of the IEEE International Conference on Prognostics and Health Management, 2012: 1-8.
22	HU Y, PALME T, FINK O. Deep health indicator extraction: a method based on auto-encoders and extreme learning machines[C]//Proc.of the Annual Conference of the Prognostics and Health Management Society, 2016: 1-7.

工况	转速/rpm	载荷/N
1	1 800	4 000
2	1 650	4 200
3	1 500	5 000

数据集	工况1	工况2	工况3
训练集	Bearing1_1	Bearing2_1	Bearing3_1
训练集	Bearing1_2	Bearing2_2	Bearing3_2
测试集	Bearing1_3	Bearing2_3	Bearing3_3
	Bearing1_4	Bearing2_4
	Bearing1_5	Bearing2_5
	Bearing1_6	Bearing2_6
	Bearing1_7	Bearing2_7

网络层	参数
输入层	尺寸:[100×100],通道数:3
卷积层1	卷积核尺寸:[11×11],步长:4,通道数:96
池化层1	卷积核尺寸:[3×3],步长:2,通道数:96
卷积层2	卷积核尺寸:[5×5],步长:1,通道数:256
池化层2	卷积核尺寸:[3×3],步长:2,通道数:256
卷积层3	卷积核尺寸:[3×3],步长:1,通道数:384
卷积层4	卷积核尺寸:[3×3],步长:1,通道数:384
卷积层5	卷积核尺寸:[3×3],步长:1,通道数:256
池化层5	卷积核尺寸:[3×3],步长:2,通道数:256
多尺度展平层	通道数:7 424(6 400+1 024)
全连接层1	通道数:4 096
全连接层2	通道数:1 000
输出层	通道数:1

轴承编号	RMS		PCA		ELM-AE		SDAE-SOM		MULCNN-HI
轴承编号	Tre	Mon	Tre	Mon	Tre	Mon	Tre	Mon	Tre	Mon
1_3	0.77	0.14	0.78	0.15	0.77	0.14	0.67	0.12	0.96	0.83
1_4	0.32	0.16	0.32	0.15	0.31	0.15	0.41	0.17	0.97	0.76
1_5	0.16	0.12	0.23	0.11	0.28	0.12	0.26	0.14	0.91	0.75
1_6	0.10	0.10	0.16	0.11	0.18	0.11	0.31	0.13	0.84	0.43
1_7	0.23	0.11	0.33	0.13	0.34	0.10	0.35	0.14	0.94	0.81

[1]	宋爽, 张悦, 张琳娜, 岑翼刚, 李浥东. 基于深度学习的轻量化目标检测算法[J]. 系统工程与电子技术, 2022, 44(9): 2716-2725.
[2]	韩啸, 陈世文, 陈蒙, 杨锦程. 基于互易点学习的LPI信号开集识别[J]. 系统工程与电子技术, 2022, 44(9): 2752-2759.
[3]	王彩云, 吴钇达, 王佳宁, 马璐, 赵焕玥. 基于改进的CNN和数据增强的SAR目标识别[J]. 系统工程与电子技术, 2022, 44(8): 2483-2487.
[4]	张立民, 谭凯文, 闫文君, 张聿远. 基于多级跳线残差网络的雷达辐射源识别[J]. 系统工程与电子技术, 2022, 44(7): 2148-2156.
[5]	金国栋, 薛远亮, 谭力宁, 许剑锟. 基于孪生神经网络的目标跟踪算法进展研究[J]. 系统工程与电子技术, 2022, 44(6): 1805-1822.
[6]	韦娟, 杨皇卫, 宁方立. 基于NMF与CNN联合优化的声学场景分类[J]. 系统工程与电子技术, 2022, 44(5): 1433-1438.
[7]	赵晓枫, 徐叶斌, 吴飞, 牛家辉, 蔡伟, 张志利. 基于全局感知机制的地面红外目标检测方法[J]. 系统工程与电子技术, 2022, 44(5): 1461-1467.
[8]	邹虹, 白陈阳, 何鹏, 崔亚平, 王汝言, 吴大鹏. 基于分布式深度学习的边缘服务放置策略[J]. 系统工程与电子技术, 2022, 44(5): 1728-1737.
[9]	陈冬, 句彦伟. 基于语义分割实现的SAR图像舰船目标检测[J]. 系统工程与电子技术, 2022, 44(4): 1195-1201.
[10]	方伟, 王玉, 闫文君, 林冲. 基于神经网络的符号化飞行动作识别[J]. 系统工程与电子技术, 2022, 44(3): 737-745.
[11]	孙晶明, 虞盛康, 孙俊. 基于深度学习的HRRP识别姿态敏感性分析[J]. 系统工程与电子技术, 2022, 44(3): 802-807.
[12]	刘恒燕, 张立民, 闫文君, 钟兆根, 凌青, 梁晓军. 基于WBP-CNN算法的LDPC译码[J]. 系统工程与电子技术, 2022, 44(3): 1030-1035.
[13]	邵凯, 朱苗苗, 王光宇. 基于生成对抗与卷积神经网络的调制识别方法[J]. 系统工程与电子技术, 2022, 44(3): 1036-1043.
[14]	张玺, 金正猛, 姜亚琴. 融合深度图像先验的全变差图像着色算法[J]. 系统工程与电子技术, 2022, 44(2): 385-393.
[15]	姚云翔, 陈莹. 注意力机制下双模态交互融合的目标跟踪网络[J]. 系统工程与电子技术, 2022, 44(2): 410-419.