基于多频带特征图和改进SqueezeNet的滚动轴承故障诊断

doi:10.13229/j.cnki.jdxbgxb.20240710

摘要/Abstract

摘要：

针对目前基于深度学习的故障诊断方法普遍存在模型参数量大和诊断时间长、在噪声环境下诊断性能也会大打折扣的问题，提出了一种快速、轻量的智能化诊断模型进行滚动轴承故障诊断。首先，利用鱼鹰优化算法（OOA）优化变分模态分解（VMD）的参数，进而基于固有模态函数（IMF）分量设计一种多频带灰度特征图；然后，基于有效注意力通道（ECA）模块设计一种残差注意力机制（RAM）模块，并集成到SqueezeNet模型中；最后，使用K最近邻（KNN）代替Softmax函数对故障进行识别与分类，建立了RSqueezeNet-KNN模型。两组实验结果表明：在噪声环境下，该模型对比其他方法能够实现轻量化应用，具有优秀的诊断性能。

关键词: 轴承故障诊断, 灰度特征图, SqueezeNet模型, 注意力机制, 轻量化

Abstract:

At present， fault diagnosis methods based on deep learning generally have the problems of large model parameters and long diagnosis times， and the diagnostic performance will be greatly reduced in noisy environments. This paper proposes a fast and lightweight intelligent diagnosis model for rolling bearing fault diagnosis. Firstly， the parameters of the variational mode decomposition （VMD） are optimized using the osprey optimization algorithm （OOA） to design a unique multi-frequency band grayscale feature map based on the intrinsic mode function （IMF） component. Then a residual attention mechanism module （RAM） is designed based on the efficient channel attention （ECA） module， which is integrated into the SqueezeNet model， and the K-nearest neighbor （KNN） method is used instead of the Softmax function to identify and classify the faults， and the RSqueezeNet-KNN model is established. Experimental results on two bearing datasets show that the model is able to achieve lightweight applications with excellent diagnostic performance compared to other methods in noisy environments.

Key words: bearing fault diagnosis, grayscale feature map, SqueezeNet model, attention mechanism, light-weighting

中图分类号:

TH17

冯志刚,任梦媛,董冰,于明月. 基于多频带特征图和改进SqueezeNet的滚动轴承故障诊断[J]. 吉林大学学报(工学版), 2026, 56(1): 96-108.

Zhi-gang FENG,Meng-yuan REN,Bing DONG,Ming-yue YU. Rolling bearing fault diagnosis based on multi-band feature map and improved SqueezeNet[J]. Journal of Jilin University(Engineering and Technology Edition), 2026, 56(1): 96-108.

图/表 26

图1

图2

图3

图4

图5

图6

图7

图8

表1

图9

图10

表2

实验一中[K,α]最佳参数组合"

标签	$K$	$α$
No1	7	2 000
Ba1	7	1 994
Ba2	6	1 999
Ba3	6	1 845
In1	7	1 991
In2	7	1 998
In3	7	1 996
Ou1	6	1 940
Ou2	7	2 000
Ou3	7	1 995

表2

图11

图12

表3

表4

表5

图13

图14

图15

表6

表7

实验二中[K,α]的最佳参数组合"

标签	$K$	$α$
Norm	7	1 999
Ball	7	1 934
Inner	7	1 991
Outer	7	945

表7

图16

图17

表8

图18

参考文献 25

[1]	Bai R, Xu Q, Meng Z, et al. Rolling bearing fault diagnosis based on multi-channel convolution neural network and multi-scale clipping fusion data augmentation[J]. Measurement, 2021, 184: No.109885.
[2]	栾孝驰, 佟鑫宇, 沙云东, 等. 基于振动与声发射敏感参数识别的主轴承故障诊断方法[J]. 推进技术, 2024, 45(12): 269-281.
	Luan Xiao-chi, Tong Xin-yu, Sha Yun-dong, et al. Main bearing fault diagnosis method based on vibration and acoustic emission snsitive parameter recognition[J]. Journal of Propulsion Technology,2024, 45(12):269-281.
[3]	谷玉海, 朱腾腾, 饶文军, 等.基于EMD二值化图像和CNN的滚动轴承故障诊断[J]. 振动、测试与诊断, 2021, 41(1): 105-113, 203.
	Gu Yu-hai, Zhu Teng-teng, Rao Wen-jun, et al. Fault diagnosis for rolling bearing based on EMD binarization image and CNN[J]. Journal of Vibration, Measurement & Diagnosis,2021,41(1):105-113, 203.
[4]	Dragomiretskiy K, Zosso D. Variational mode decomposition[J]. IEEE Transactions on Signal Processing, 2013, 62(3): 531-544.
[5]	赵德宏, 李永利. 基于GA-VMD分解与支持向量机的刀具故障诊断研究[J].沈阳建筑大学学报: 自然科学版, 2024, 40(2): 361-371.
	Zhao De-hong, Li Yong-li. Research on tool fault diagnosis based on GA-VMD decomposition and support vector machine[J]. Journal of Shenyang Jianzhu University(Natural Science), 2024, 40(2): 361-371.
[6]	余萍, 赵康, 曹洁. 基于优化A-BiLSTM的滚动轴承故障诊断[J]. 吉林大学学报: 工学版, 2024, 54(8): 2156-2166.
	Yu Ping, Zhao Kang, Cao Jie. Rolling bearing fault diagnosis based on optimized A-BiLSTM[J]. Journal of Jilin University(Engineering and Technology Edition), 2024, 54(8): 2156-2166.
[7]	杨战社, 孔晨再, 荣相, 等. 基于EEMD能量熵与ANN的矿用异步电机故障诊断[J]. 微电机, 2021, 54(8): 23-27, 61.
	Yang Zhan-she, Kong Chen-zai, Rong Xiang, et al. Fault diagnosis of mine asynchronous motor based on EEMD energy entropy and ANN[J]. Micromotors,2021, 54(8): 23-27, 61.
[8]	Pandya D H, Upadhyay S H, Harsha S P. Fault diagnosis of rolling element bearing with intrinsic mode function of acoustic emission data using APF-KNN[J]. Expert Systems with Applications, 2013, 40(10): 4137-4145.
[9]	Che C, Wang H, Xiong M, et al. Few-shot fault diagnosis of rolling bearing under variable working conditions based on ensemble meta-learning[J]. Digital Signal Processing, 2022, 131: No. 103777.
[10]	Gu J, Peng Y, Lu H, et al. A novel fault diagnosis method of rotating machinery via VMD, CWT and improved CNN[J]. Measurement, 2022, 200:No. 111635.
[11]	Wang Z, Zhao W, Du W, et al. Data-driven fault diagnosis method based on the conversion of erosion operation signals into images and convolutional neural network[J]. Process Safety and Environmental Protection, 2021, 149: 591-601.
[12]	Iandola F N, Han S, Moskewicz M W, et al. SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and<0.5 MB model size[J/OL].[2024-06-12]. .
[13]	Ren L, Jiang L, Li C. Label confidence-based noise correction for crowdsourcing[J]. Engineering Applications of Artificial Intelligence, 2023, 117: No.105624.
[14]	Dehghani M, Trojovský P. Osprey optimization algorithm: a new bio-inspired metaheuristic algorithm for solving engineering optimization problems[J]. Frontiers in Mechanical Engineering, 2023, 8:No. 1126450.
[15]	Lin G, Lin A. Modified multiscale sample entropy and cross-sample entropy based on horizontal visibility graph[J]. Chaos, Solitons & Fractals, 2022, 165:No. 112802.
[16]	Han G, Zhang M, Wu W, et al. Improved U-Net based insulator image segmentation method based on attention mechanism[J]. Energy Reports, 2021, 7: 210-217.
[17]	Smith W A, Randall R B. Rolling element bearing diagnostics using the Case Western Reserve University data: a benchmark study[J]. Mechanical Systems and Signal Processing, 2015, 64/65: 100-131.
[18]	Wu Q L, Lin H X. Short-term wind speed forecasting based on hybrid variational mode decomposition and least squares support vector machine optimized by bat algorithm model[J]. Sustainability, 2019, 11(3): No.652.
[19]	Diao X, Jiang J, Shen G, et al. An improved variational mode decomposition method based on particle swarm optimization for leak detection of liquid pipelines[J]. Mechanical Systems and Signal Processing, 2020, 143: No.106787.
[20]	Nadirgil O. Carbon price prediction using multiple hybrid machine learning models optimized by genetic algorithm[J]. Journal of Environmental Management, 2023, 342: No.118061.
[21]	Wang H, Wu F, Zhang L. Application of variational mode decomposition optimized with improved whale optimization algorithm in bearing failure diagnosis[J]. Alexandria Engineering Journal, 2021, 60(5): 4689-4699.
[22]	Liang P, Wang W, Yuan X, et al. Intelligent fault diagnosis of rolling bearing based on wavelet transform and improved ResNet under noisy labels and environment[J]. Engineering Applications of Artificial Intelligence, 2022, 115: No.105269.
[23]	Hao S, Ge F X, Li Y, et al. Multisensor bearing fault diagnosis based on one-dimensional convolutional long short-term memory networks[J]. Measurement, 2020, 159: No.107802.
[24]	Xu Z, Li C, Yang Y. Fault diagnosis of rolling bearing of wind turbines based on the variational mode decomposition and deep convolutional neural networks[J]. Applied Soft Computing, 2020, 95: No.106515.
[25]	Feng Z, Wang S, Yu M. A fault diagnosis for rolling bearing based on multilevel denoising method and improved deep residual network[J]. Digital Signal Processing, 2023, 140: No.104106.

相关文章 15

[1]	周求湛,李新萌,沈皓庆子,武慧南,李媛媛,荣静,胡春华,刘萍萍. 基于注意力机制的不平衡数据的非侵入式负荷分解[J]. 吉林大学学报(工学版), 2026, 56(1): 239-246.
[2]	霍震,金立生,华强,贺阳. 基于边缘特征引导的智能汽车语义分割方法[J]. 吉林大学学报(工学版), 2025, 55(9): 3032-3041.
[3]	刘长钊,宋健,李峥琪,张铁,王磊. 考虑动力学性能的高速薄壁齿轮多目标优化[J]. 吉林大学学报(工学版), 2025, 55(8): 2487-2500.
[4]	庄珊娜,王君帅,白晶,杜京瑾,王正友. 基于三维卷积与自注意力机制的视频行人重识别[J]. 吉林大学学报(工学版), 2025, 55(7): 2409-2417.
[5]	冯志刚,王首起,于明月. 基于变分模态提取及轻量级网络的滚动轴承故障诊断[J]. 吉林大学学报(工学版), 2025, 55(6): 1883-1891.
[6]	薛雅丽,俞潼安,崔闪,周李尊. 基于级联嵌套U-Net的红外小目标检测[J]. 吉林大学学报(工学版), 2025, 55(5): 1714-1721.
[7]	才华,王玉瑶,付强,马智勇,王伟刚,张晨洁. 基于注意力机制和特征融合的语义分割网络[J]. 吉林大学学报(工学版), 2025, 55(4): 1384-1395.
[8]	张河山,范梦伟,谭鑫,郑展骥,寇立明,徐进. 基于改进YOLOX的无人机航拍图像密集小目标车辆检测[J]. 吉林大学学报(工学版), 2025, 55(4): 1307-1318.
[9]	张兰芳,李根泽,刘婷宇,余博. 局部多车影响下跟驰行为机理及建模[J]. 吉林大学学报(工学版), 2025, 55(3): 963-973.
[10]	李扬,李现国,苗长云,徐晟. 基于双分支通道先验和Retinex的低照度图像增强算法[J]. 吉林大学学报(工学版), 2025, 55(3): 1028-1036.
[11]	程德强,刘规,寇旗旗,张剑英,江鹤. 基于自适应大核注意力的轻量级图像超分辨率网络[J]. 吉林大学学报(工学版), 2025, 55(3): 1015-1027.
[12]	董华松,连远锋. 海量数字媒体视频无损转码重压缩的轻量化检测算法[J]. 吉林大学学报(工学版), 2025, 55(2): 741-747.
[13]	李云红,王梅,苏雪平,李丽敏,张富星,郝特吉. 结合注意力与上下文融合的遥感图像道路提取[J]. 吉林大学学报(工学版), 2025, 55(12): 4034-4044.
[14]	王祥,谭国真,彭衍飞,任浩,李健平. 基于语言推理和认知记忆的自动驾驶决策模型[J]. 吉林大学学报(工学版), 2025, 55(12): 3918-3927.
[15]	彭铎,刘明硕,谢堃. 观测站参数误差下联合多特征融合注意力机制的TDOA/FDOA多机无源定位算法[J]. 吉林大学学报(工学版), 2025, 55(11): 3751-3761.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed

轴承状态	直径/英寸	标签	数据类型	样本数
正常	—	Nor	1 024	200
滚动体故障	0.007	Ba1	1 024	200
	0.014	Ba2	1 024	200
	0.021	Ba3	1 024	200
内圈故障	0.007	In1	1 024	200
	0.014	In2	1 024	200
	0.021	In3	1 024	200
外圈故障	0.007	Ou1	1 024	200
	0.014	Ou2	1 024	200
	0.021	Ou3	1 024	200

超参数组合	正确率/%	训练时间/s
A	99.01	145.6
B	99.15	101.5
C	99.08	92.8
D	99.43	150.6
E	99.70	85.6
F	99.58	90.5
G	96.02	149.5
H	95.10	105.1
I	88.78	89.0

模型	参数/MB	训练时间/s
RSqueezeNet-KNN	0.015	85.849
RSqueezeNet	0.015	85.6
SqueezeNet	0.031	95.2
AlexNet	24.723	143.2
ResNet	0.041	121.7
WT-IResNet^［22］	0.132	661
WT-ResNet^［22］	0.148	278

轴承状态	标签	数据长度	样本数	转速/（r·min^-1）
正常	Norm	1 024	800	1 500
滚动体故障	Ball	1 024	800	1 500
内圈故障	Inner	1 024	800	1 500
外圈故障	Outer	1 024	800	1 500