多图卷积网络模型(Multi-Graph Convolutional Network, M-GCN)可将图像转为特征向量,并可利用图卷积操作增强同类节点聚集。由于受到现场空间和经济条件限制,无法有效地采集液压泵充足故障样本,导致小样本问题;当引入M-GCN模型对液压泵的故障进行诊断时,该模型特征表达存在区分度不足和信息单一等问题。因此,本文提出了一种改进多图卷积网络模型,即MMH-GCN模型。首先,为解决模型特征提取区分度不足问题,引入掩码自编码器(Masked Autoencoder, MAE)降低编码维度并提取关键图像特征,提升模型的小样本诊断精度;然后,为解决模型特征信息单一问题,引入异构图注意力网络(Heterogeneous Graph Attention Network, HAN)提取更丰富和全面的图结构数据特征,以提升模型的小样本诊断精度和效率。通过液压泵实测故障实验验证分析可知,本文所提MMH-GCN模型较原模型具有明显的高效性和优越性,在诊断精度和效率方面分别提升了12.14%和14.63%。
Abstract
Images can be converted into feature vectors by multi-graph convolutional network (M-GCN), and the aggregation of similar nodes can be enhanced by graph convolution operation. Due to the limitation of field space and economic conditions, it is impossible to collect sufficient fault samples of hydraulic pump, resulting the problem of small sample. When M-GCN is introduced to diagnose fault of hydraulic pump, there are problems such as insufficient discrimination and single information in the feature expression. Therefore, an improved multi-graph convolutional network, namely MMH-GCN, was proposed in this paper. In order to solve the problem of insufficient discrimination of model feature extraction, masked autoencoder (MAE) was introduced to reduce the encoding dimension and extract key image features, so as to improve diagnostic accuracy based on small sample size. For solving the problem of single feature information, the heterogeneous graph attention network (HAN) was introduced to extract much abundant and comprehensive features of graph structure data, so as to improve diagnostic accuracy and efficiency based on small sample size. Through the measured fault experimental verification and analysis of hydraulic pump, it can be seen that the MMH-GCN proposed in this paper has obvious efficiency and superiority compared with the original M-GCN, and the diagnostic accuracy and efficiency are increased by 12.14% and 14.63% respectively.
关键词
多图卷积网络 /
掩码自编码器 /
异构图注意力网络 /
小样本
{{custom_keyword}} /
Key words
multi-graph convolutional network /
masked autoencoder /
heterogeneous graph attention network /
small sample
{{custom_keyword}} /
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] 王浩任, 黄亦翔, 赵帅, 等. 基于小波包和拉普拉斯特征值映射的柱塞泵健康评估方法[J].振动与冲击, 2017, 36(22): 45-50.
WANG Haoren, HUANG Yixiang, ZHAO Shuai, et al. Health assessment for piston pump based on wpd and le[J]. Journal of Vibration and Shock, 2017, 36(22): 45-50.
[2] GUO J , SI Z , XIANG J. Cycle kurtosis entropy guided symplectic geometry mode decomposition for detecting faults in rotating machinery [J]. ISA transactions, 2023, 138: 546-561.
[3] 姜万录, 赵亚鹏, 张淑清, 等. 精细复合多尺度波动散布熵在液压泵故障诊断中的应用[J]. 振动与冲击, 2022, 41(08): 7-16.
JIANG Wanlu, ZHAO Yapeng, ZHANG Shuqing, et al. Application of fine composite multi-scale wave dispersion entropy in fault diagnosis of hydraulic pump[J]. Journal of Vibration and Shock, 2022, 41(08): 7-16.
[4] FENG J, BAO S Y, XU X B, et al. Rotating machinery fault diagnosis based on feature extraction via an unsupervised graph neural network[J]. Applied Intelligence, 2023, 53(18): 21211-21226.
[5] DING A, QIN Y, WANG B, et al. Evolvable graph neural network for system-level incremental fault diagnosis of train transmission systems[J]. Mechanical Systems and Signal Processing, 2024, 210: 111175.
[6] ZHANG Z W, WU L F. Graph neural network-based bearing fault diagnosis using granger causality test[J]. Expert Systems With Applications, 2024, 242: 122827.
[7] THOMAS N. KIPF, MAX WELLING. Semi-supervised classification with graph convolutional networks[J]. IEICE transactions on fundamentals of electronics, communications and computer sciences, 2016, 160: 902907.
[8] CHEN Z, XU J, PENG T, et al. Graph Convolutional Network-based method for fault diagnosis using a hybrid of measurement and prior knowledge[J]. IEEE Transactions on Cybernetics, 2021: 1-13.
[9] CHAUDHURI A, TALUKDAR J, JUNG J, et al. Fault-Criticality assessment for ai accelerators using graph convolutional networks[J]. 2021 Design, Automation & Test in Europe Conference & Exhibition (DATE), 2021: 1596-1599.
[10] LIAO W, YANG D, WANG Y, et al. Fault diagnosis of power transformers using graph convolutional network[J]. CSEE journal of power and energy systems, 2021, 7(2): 241-249.
[11] 陈杰虎, 汪西莉. 多图卷积网络的遥感图像小样本分类[J]. 遥感学报, 2022, 26(10): 2029-2042.
CHEN Jiehu, Wang Xili. Remote sensing image few-shot classification based on multi-graph convolutional network[J]. Journal of Remote Sensing, 2022, 26(10): 2029-2042.
[12] LIU Y Y, COLIN P, et al. Multiresolution convolutional autoencoders[J]. Journal of Computational Physics, 2023, 474: 111801.
[13] NIU M G, JIANG H K, WU Z H, et al. An enhanced sparse autoencoder for machinery interpretable fault diagnosis[J]. Measurement Science and Technology, 2024, 35(5): 055108.
[14] LOPES LVANDRO O, et al. Effective network intrusion detection via representation learning: A denoising autoencoder approach[J]. Computer Communications, 2022, 194: 55-65.
[15] HE K, CHEN X, XIE S, et al. Masked autoencoders are scalable vision learners[J]. Computer Vision and Pattern Recognition, 2021: 06377.
[16] WU Y J, ZHOU J T. A neighborhood-aware graph self-attention mechanism-based pre-training model for knowledge graph reasoning[J]. Information Sciences, 2023, 647: 119473.
[17] WANG B W, WANG J S. ST-MGAT: Spatio-temporal multi-head graph attention network for traffic prediction[J]. Physical A: Statistical Mechanics and its Applications, 2022, 603: 127762.
[18] WANG X, JI H Y, SHI C, et al. Heterogeneous graph attention network[J]. IEICE Transactions on Fundamentals of Electronics, Communications and Computer Science, 2019: 07293.
[19] FRANCO S, MARCO G, CHUNG A T, et al. The graph neural network model[J]. IEEE transactions on neural networks, 2009, 20(1): 61-80.
{{custom_fnGroup.title_cn}}
脚注
{{custom_fn.content}}