改进胶囊网络优化分层卷积的亚健康识别算法

doi:10.3778/j.issn.1673-9418.2004017

摘要/Abstract

摘要：

针对传统卷积神经网络（CNN）为获得高准确率不断堆叠卷积层、池化层致使模型结构复杂、训练时间长且数据处理方式单一的问题，提出改进胶囊网络优化分层卷积的亚健康识别算法。首先，对原始振动数据进行小波降噪和小波包降噪两种数据处理，更好地保留原始信号中对亚健康识别有用的信息；其次，CNN采用分层卷积的思想，并行3个不同尺度的卷积核，多角度地进行特征提取；最后,将卷积核提取的特征输入到剪枝策略的胶囊网络中进行亚健康识别，改进的胶囊网络在保证准确率的同时加快亚健康识别时间，解决CNN结构过于复杂以及识别效果不佳的问题。实验结果表明，提出算法识别准确率高且识别时间较少。

关键词: 亚健康识别, 卷积神经网络（CNN）, 胶囊网络, 小波降噪, 小波包降噪

Abstract:

Aiming at the problem that traditional convolutional neural network (CNN) continuously stacks convo-lutional layers and pooling layers in order to obtain high accuracy, resulting in complicated model structure, long training time, and single data processing method, a optimized layered convolutional sub-health recognition algorithm of improved capsule network is proposed. Firstly, the original vibration data are transformed by wavelet denoising and wavelet packet denoising to better retain the useful information in the original signal for sub-health recognition. Secondly, CNN adopts the idea of layered convolution, parallelizes three convolution kernels of different scales, and carries on multi-angle feature extraction. Finally, the features extracted by the convolution kernels are input into the improved capsule network with pruning strategy for sub-health recognition. The improved capsule network can not only guarantee the accuracy, but also accelerate the sub-health recognition time, thus the problems of too compli-cated CNN structure and poor recognition effect are solved. Experimental results show that the proposed algorithm has high recognition accuracy and less recognition time.

Key words: sub-health recognition, convolutional neural network (CNN), capsule network, wavelet denoising, wavelet packet denoising

张利, 邱存月, 张凯鑫, 张大波, 罗浩. 改进胶囊网络优化分层卷积的亚健康识别算法[J]. 计算机科学与探索, 2021, 15(4): 712-722.

ZHANG Li, QIU Cunyue, ZHANG Kaixin, ZHANG Dabo, LUO Hao. Optimized Layered Convolutional Sub-health Recognition Algorithm of Improved Capsule Network[J]. Journal of Frontiers of Computer Science and Technology, 2021, 15(4): 712-722.

参考文献

[1] WANG Z Y, YAO L G, CAI Y W. Rolling bearing fault dia-gnosis using generalized refined composite multiscale sample entropy and optimized support vector machine[J]. Measure-ment, 2020, 156(3): 107574.
[2] LI J M, YAO X F, WANG X D. Multiscale local features lear-ning based on BP neural network for rolling bearing intelli-gent fault diagnosis[J]. Measurement, 2020, 153(1): 107419.
[3] LI X, ZHANG W, DING Q. Understanding and improving deep learning-based rolling bearing fault diagnosis with atten-tion mechanism[J]. Signal Processing, 2019, 161(8): 136-154.
[4] LU S L, HE Q, ZHAO J W. Bearing fault diagnosis of a per-manent magnet synchronous motor via a fast and online order analysis method in an embedded system[J]. Mechani-cal Systems and Signal Processing, 2018, 113(12): 36-49.
[5] LIU C, CHENG G, LIU B. Bearing fault diagnosis method with unknown variable speed on multi-curve extraction and selection[J]. Measurement, 2020, 153(1): 107437.
[6] HE M, HE D. A new hybrid deep signal processing approach for bearing fault diagnosis using vibration signals[J]. Neuro-computing, 2019, 24(4): 107523-107531.
[7] GU Y K, ZENG L, QIU G Q. Bearing fault diagnosis with varying conditions using angular domain resampling tech-nology, SDP and DCNN[J].Measurement, 2020, 156(3): 107616.
[8] HOANG D T, KANG H J. Rolling element bearing fault diagnosis using convolutional neural network and vibration image[J]. Cognitive Systems Research, 2019, 53(1): 42-50.
[9] ISLAM M, KIM J M. Automated bearing fault diagnosis scheme using 2D representation of wavelet packet transform and deep convolutional neural network[J]. Computers in Indu-stry, 2019, 106(4): 142-153.
[10] SABOUR S, FROSST N, HINTON G E. Dynamic routing between capsules[C]//Proceedings of the Advances in Neural Information Processing Systems, Long Beach, Dec 4-9, 2017. Red Hook: Curran Associates, 2017: 3856-3866.
[11] ZHU Z Y, PENG G L, CHEN Y H. A convolutional neural net-work based on a capsule network with strong generaliza-tion for bearing fault diagnosis[J]. Neurocomputing, 2019, 323(1): 62-75.
[12] CHEN T Y, WANG Z H, YANG X. A deep capsule neural net-work with stochastic delta rule for bearing fault diagnosis on raw vibration signals[J]. Measurement, 2019, 148(12): 106857.
[13] ZHANG W. Study on bearing fault diagnosis algorithm based on convolutional neural network[D]. Harbin: Harbin Institute of Technology, 2017.
张伟. 基于卷积神经网络的轴承故障诊断算法[D]. 哈尔滨: 哈尔滨工业大学, 2017.
[14] ZHANG J Q, SUN Y, GUO Y. A new bearing fault diagnosis method based on modified convolutional neural networks[J]. Chinese Journal of Aeronautics, 2020, 33(2): 439-447.
[15] LU C, WANG Z, ZHOU B. Intelligent fault diagnosis of roll-ing bearing using hierarchical convolutional network based health state classification[J]. Advanced Engineering Infor-matics, 2017, 32(4): 139-151.
[16] CHEN Z Y, GRYLLIAS K, LI W H. Mechanical fault diagnosis using convolutional neural networks and extreme learning machine[J]. Mechanical Systems and Signal Processing, 2019, 133(11): 106272.
[17] SUN C, YANG Z H, WANG L. Attention guided capsule net-works for chemical-protein interaction extraction[J]. Journal of Biomedical Informatics, 2020, 103(3): 103392.
[18] ZHENG Y P, LI G Y, LI Y. Survey of application of deep lear-ning in image recognition[J]. Computer Engineering and App-lications, 2019, 55(12): 20-36.
郑远攀, 李广阳, 李晔. 深度学习在图像识别中的应用研究综述[J]. 计算机工程与应用, 2019, 55(12): 20-36.
[19] PENG D L, ZHANG D D, LIU C，et al. BG-SAC: entity relationship classification model based on self-attention sup-ported capsule networks[J]. Applied Soft Computing, 2020, 91(6): 106186-106194.
[20] LEI K, FU Q A, YANG M. Tag recommendation by text classi-fication with attention-based capsule network[J]. Neurocom-puting, 2020, 28(1): 1016-1022.
[21] CHEN H, LI G Y, QI R H, et al. Capsule network??s applica-tion in knowledge graph completion[J]. Computer Engineer-ing and Applications, 2020, 56(8): 110-116.
陈恒, 李冠宇, 祁瑞华, 等. 胶囊网络在知识图谱补全中的应用[J]. 计算机工程与应用, 2020, 56(8): 110-116.
[22] LIU J, LI Y L, LIN M. Research of short text multi-intent detection with capsule network[J]. Journal of Frontiers of Computer Science and Technology, 2020, 14(10): 1735-1743.
刘娇, 李艳玲, 林民. 胶囊网络用于短文本多意图识别的研究[J]. 计算机科学与探索, 2020, 14(10): 1735-1743.
[23] GUO L Q, CAO X, LIU L. Dual-tree biquaternion wavelet transform and its application to color image fusion[J]. Signal Processing, 2020, 171(6): 107513.
[24] LEE S J, KIM J H. Discrete wavelet transform-based denois-ing technique for advance state-of-charge estimator of a lithium-ion battery in electric vehicles[J]. Energy, 2015, 83(1): 462-473.
[25] JIANG F B, DONG L, DAI Q W. Using wavelet packet denois-ing and ANFIS networks based on COSFLA optimization for electrical resistivity imaging inversion[J]. Fuzzy Sets and Systems, 2018, 337(1): 93-112.
[26] ZHANG X K, SUN Y, WANG Y. A novel effective and effi-cient capsule network via bottleneck residual block and automated gradual pruning[J]. Computers & Electrical Engi-neering, 2019, 80(12): 106481.
[27] LI X, YANG Y, PAN H Y. A novel deep stacking least squares support vector machine for rolling bearing fault diagnosis[J]. Computers in Industry, 2019, 110(9): 3647-3656.