针对滚动轴承故障诊断模型在变工况和环境噪声干扰下诊断精度降低的问题，提出一种基于双注意力机制的多尺度卷积网络(Dual attention and Multi-scale convolutional networks，DAMSCN)与改进的双向门控循环单元(Bidirectional gated recurrent unit, BiGRU)组成的故障诊断模型 DAMSCN-BiGRU。首先，多尺度特征融合模块(Multi-scale feature fusion module, MSF)使用不同大小的卷积核，获得多种感受野，从而提取到轴承原始振动信号的多尺度特征信息，并根据重要性对其进行自适应融合，然后利用通道注意力和空间注意力组成的双注意力模块(Dual attention module, DAM)对多尺度特征进行重新标定，分配注意力权重，削弱融合特征中的冗余特征；然后，增加注意力层和利用分段激活改进BiGRU进而挖掘信号的时域特征，以提高轴承故障诊断的性能；最后，通过Softmax层完成对不同故障的分类。实验结果表明，与其他智能诊断模型相比，DAMSCN-BiGRU在变工况环境下，平均诊断精度达到98.2%，在强噪声背景下仍然有着85.3%的准确率，且在不同程度的噪声强度下效果均优于其他常用模型，有利于促进滚动轴承的智能故障诊断研究和实际应用。
Abstract
Regarding the problem that the diagnosis accuracy of rolling bearing fault diagnosis model decreases under the variable working conditions and environmental noise interference, a rolling bearing fault diagnosis method (DAMSCN-BiGRU) composed of Multi-scale Convolutional Network based on Dual Attention mechanism (DAMSCN) and improved Bidirectional Gated Recurrent Unit (BiGRU) was proposed. Firstly, using multi-scale feature fusion module with different kernel sizes to obtain a variety of receptive fields and extract the multi-scale feature information of the original vibration signal of the bearing, which were fused adaptively according to their importance. And the multi-scale features were weighted and fused using a dual attention module composed of channel attention and spatial attention to weaken the redundant features in the fused features. Then, the attention layer was added and the segmented activation was used to improve BiGRU to mine the time-domain features of the signal to improve the performance of the bearing fault diagnosis. Finally, the classification of different faults was completed by Softmax layer. The experimental results show that compared with other intelligent diagnosis models, DAMSCN-BiGRU can achieve an average diagnostic accuracy of 98.2% under variable working condition and still has an accuracy of 85.3% in the strong noise background, and the effect is better than other commonly used models under different levels of noise intensity, which is beneficial to promote the research and practical application of intelligent fault diagnosis of rolling bearings.
关键词
滚动轴承；故障诊断；多尺度特征融合；双注意力机制；双向门控循环单元
{{custom_keyword}} /
Key words
rolling bearing /
fault diagnosis /
multi-scale feature fusion /
dual attention mechanism /
bidirectional gated recurrent unit
{{custom_keyword}} /
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] Hoang D T, Kang H J. A survey on deep learning based bearing fault diagnosis[J]. Neurocomputing, 2019, 335: 327-335.
[2] Song L Y, Wang H Q, Chen Peng. Vibration-based Intelligent Fault Diagnosis for Roller Bearings in Low-speed Rotating Machinery[J]. IEEE Transactions on Instrumentation and Measurement, 2018, 67(8): 1887-1899.
[3] 葛江华,刘奇,等.支持张量机与KNN-AMDM决策融合的齿轮箱故障诊断方法[J].振动工程学报,2018,31(06):1093-1101.
GE Jiang-hua, LIU Qi, XU Di et al. Support Tensor Machine and KNN-AMDM Decision Fusion for Fault Diagnosis of Gearboxes[J]. Journal of Vibration Engineering 2018, 31(06): 1093-1101.
[4] 丁嘉鑫,王振亚,姚立纲等.广义复合多尺度加权排列熵与参数优化支持向量机的滚动轴承故障诊断[J].中国机械工程,2021,32(02):147-155.
DING Jia-xin, WANG Zhen-ya, YAO LI-gang et al. Generalized composite multiscale weighted arrangement entropy and parameter optimized support vector machine for rolling bearing fault diagnosis[J]. China Mechanical Engineering, 2021, 32(02):147-155.
[5] 陈仁祥,黄鑫,胡小林等.多源信息深度融合的行星齿轮箱故障诊断方法[J].振动工程学报,2020,33(05):1094-1102.
CHEN Ren-xiang, HUANG Xin, HU Xiao-lin et al. Fault diagnosis method of planetary gearbox with deep fusion of multi-source information[J]. Journal of Vibration Engineering, 2020, 33(05):1094-1102.
[6] Lei Y G, Jia F, Lin J, et al. An Intelligent Fault Diagnosis Method Using Unsupervised Feature Learning towards Mechanical Big Data[J]. IEEE Transactions on Industrial Electronics, 2016, 63(5): 3137-3147.
[7] Shao H D, Jiang H K, Zhao H W, Wang F. A novel deep autoencoder feature learning method for rotating machinery fault diagnosis. Mechanical Systems and Signal Processing, 2017, 95: 187-204.
[8] Sun W J, Shao S Y, Zhao R, Yan R Q, et al. A sparse auto-encoder-based deep neural network approach for induction motor faults classification. Measurement 2016;89:171e8..
[9] Eren L, Ince T, Kiranyaz S. A generic intelligent bearing fault diagnosis system using compact adaptive 1D CNN classifier[J]. Journal of Signal Processing Systems, 2019, 91(2): 179-189.
[10] Gao Y, Kim C H, Kim J M. A novel hybrid deep learning method for fault diagnosis of rotating machinery based on extended WDCNN and long short-term memory[J]. Sensors, 2021, 21(19): 6614.
[11] Hao X, Zheng Y, Lu L, et al. Research on Intelligent Fault Diagnosis of Rolling Bearing Based on Improved Deep Residual Network[J]. Applied Sciences, 2021, 11(22): 10889.
[12] Smith WA, Randall RB. Rolling element bearing diagnostics using the Case Western Reserve University data: a benchmark study. Mech Syst Signal Process 2015;64e65:100e31.
[13] Hasan M J, Sohaib M, Kim J M. 1D CNN-based transfer learning model for bearing fault diagnosis under variable working conditions[C]. International Conference on Computational Intelligence, 2018: 13-23.
[14] Szegedy C, Liu W, Jia Y, et al. Going deeper with convolutions[C]. Proceedings of the IEEE conference on computer vision and pattern recognition. 2015: 1-9.
[15] Graves A. Generating sequences with recurrent neural networks, vol. 43; 2014. arXiv:13080850.
{{custom_fnGroup.title_cn}}
脚注
{{custom_fn.content}}
PDF(2335 KB)
