VAE model combined with multi-domain feature for muscle fatigue analysis

doi:10.3785/j.issn.1008-973X.2026.06.019

Journal of ZheJiang University (Engineering Science)

2026, Vol. 60

Issue (6): 1317-1328 DOI: 10.3785/j.issn.1008-973X.2026.06.019

VAE model combined with multi-domain feature for muscle fatigue analysis

Bo DONG1(

),Donghao LV1,2,*(

),Dahua YU1,Xiaowei DU3

1. School of Automation and Electrical Engineering, Inner Mongolia University of Science and Technology, Baotou 014010, China
2. School of Automation, China University of Geosciences, Wuhan 430074, China
3. Department of Physical Education, Inner Monglia University of Science and Technology, Baotou 014010, China

Download:

HTML

PDF(1447KB) HTML
Export: BibTeX | EndNote (RIS)

Abstract

A structure-smoothing regularized variational autoencoder (SSR-VAE) was proposed based on the variational autoencoder (VAE) in order to address the issue of single-feature reliance, limited capacity to capture complex fatigue evolution process, and unclear delineation of fatigue state in existing muscle fatigue analysis method. Time-domain, frequency-domain and nonlinear entropy feature were incorporated to enhance the model’s ability to capture the dynamic change in surface electromyography (sEMG) signal. The decoupling and temporal continuity of the latent variable space were optimized by introducing weighted KL divergence and trajectory smoothing regularization term into the loss function. A correlation-based selection mechanism was used to extract the main fatigue factor from the latent space, which effectively represented the muscle fatigue state. The experimental results show that SSR-VAE outperforms the VAE model in reconstruction performance, and the extracted main fatigue factor can be used to more clearly distinguish different fatigue stage. SSR-VAE achieved a maximum accuracy of 96.211% in muscle fatigue state classification, significantly outperforming other comparison method.

Key words： muscle fatigue surface electromyography signal variational autoencoder (VAE) latent variable feature fusion

Received: 01 July 2025 Published: 06 May 2026

CLC:

TN 911

Fund: 内蒙古自治区自然科学基金资助项目(2024MS06024)；内蒙古自治区一流学科科研专项资助项目(YLXKZX-NKD-020)；内蒙古自治区直属高校基本科研业务费资助项目(2023QNJS194).

Corresponding Authors: Donghao LV E-mail: 2023023236@stu.imust.edu.cn;wsldh2016957@imust.edu.cn

	Service
	E-mail this article
	Add to my bookshelf
	Add to citation manager
	E-mail Alert
	RSS
	Articles by authors
	Bo DONG
	Donghao LV
	Dahua YU
	Xiaowei DU

Cite this article:

Bo DONG,Donghao LV,Dahua YU,Xiaowei DU. VAE model combined with multi-domain feature for muscle fatigue analysis. Journal of ZheJiang University (Engineering Science), 2026, 60(6): 1317-1328.

URL:

https://www.zjujournals.com/eng/10.3785/j.issn.1008-973X.2026.06.019 OR https://www.zjujournals.com/eng/Y2026/V60/I6/1317

融合多域特征的VAE模型在肌肉疲劳分析中的应用

针对现有肌肉疲劳分析方法中存在的特征单一、对复杂疲劳演化过程捕捉能力不足以及疲劳状态界定不清等问题，基于变分自编码器(VAE)提出结构-平滑正则化变分自编码器(SSR-VAE). 结合时域、频域和非线性熵3类特征输入，增强对表面肌电(sEMG)信号动态变化的捕捉能力. 通过在损失函数中引入加权KL散度和轨迹平滑正则化项，优化了潜变量空间的解耦性和时序连续性. 利用相关性筛选机制，从潜变量空间中提取主疲劳因子，有效表征肌肉疲劳状态. 实验结果表明，与VAE模型相比，SSR-VAE在重构性能方面表现更优，利用提取的主疲劳因子，能够更清晰地划分疲劳阶段. SSR-VAE在肌肉疲劳状态分类中的最高准确率达到96.211%，明显优于其他对比方法.

关键词： 肌肉疲劳, 表面肌电信号, 变分自编码器(VAE), 潜变量, 特征融合

Fig.1 SSR-VAE model

Tab.1 Optimization result of KL weighted parameter

Fig.2 Feature extraction result of sEMG signal

Fig.3 Convergence curve of two models

Fig.4 Feature reconstruction result of variational autoencoder model

Fig.5 Feature reconstruction results of SSR-VAE

Fig.6 Correlation between latent variable and feature

Fig.7 Main fatigue factor and clustering result

Tab.2 Quantification result of muscle fatigue in each subject

Tab.3 Result of muscle fatigue classification with different feature %

Fig.8 Confusion matrix result of muscle fatigue classification experiment

Tab.4 Result of ablation experiment


[1]	MARCO G, ALBERTO B, TAIAN V Surface EMG and muscle fatigue: multi-channel approaches to the study of myoelectric manifestations of muscle fatigue[J]. Physiological Measurement, 2017, 38 (5): R27 doi: 10.1088/1361-6579/aa60b9

[2]	YOUSIF H A, NORASMADI A R, BIN SALLEH A F, et al Assessment of muscles fatigue during 400-meters running strategies based on the surface EMG signals[J]. Journal of Biomimetics, Biomaterials and Biomedical Engineering, 2019, 42: 1- 13 doi: 10.4028/www.scientific.net/jbbbe.42.1

[3]	ZHOU Y, LI J, DONG M Prediction of actively exerted torque from ankle joint complex based on muscle synergy[J]. IEEE Transactions on Industrial Electronics, 2024, 71 (2): 1729- 1737 doi: 10.1109/TIE.2023.3257380

[4]	CHLIF M, KEOCHKERIAN D, TEMFEMO A, et al Relationship between electromyogram spectrum parameters and the tension-time index during incremental exercise in trained subjects[J]. Journal of Sports Science and Medicine, 2018, 17 (3): 509- 514

[5]	王万良, 潘杰, 王铮, 等基于双分支网络的表面肌电信号识别方法[J]. 浙江大学学报: 工学版, 2024, 58 (11): 2208- 2218 WANG Wanliang, PAN Jie, WANG Zheng, et al Recognition method of surface electromyographic signal based on two-branch network[J]. Journal of Zhejiang University: Engineering Science, 2024, 58 (11): 2208- 2218 doi: 10.3785/j.issn.1008-973X.2024.11.002

[6]	GUO H, GONG P, WANG Y, et al Complex network properties analysis of muscle fatigue based on sEMG signals[J]. IEEE Sensors Journal, 2023, 23 (4): 3859- 3869 doi: 10.1109/JSEN.2022.3233047

[7]	DUAN T, HUANG B, LI X, et al Real-time indicators and influence factors of muscle fatigue in push-type work[J]. International Journal of Industrial Ergonomics, 2020, 80: 103046 doi: 10.1016/j.ergon.2020.103046

[8]	姚贺龙, 吕东澔, 张勇, 等基于傅里叶分解方法的肌肉疲劳状态分类研究[J]. 电子测量与仪器学报, 2023, 37 (6): 48- 58 YAO Helong, LV Donghao, ZHANG Yong, et al Study of muscle fatigue state classification based on Fourier decomposition method[J]. Journal of Electronic Measurement and Instrumentation, 2023, 37 (6): 48- 58 doi: 10.13382/j.jemi.B2306358

[9]	BIRNBAUM S, SHARSHAR T, ROPERS J, et al Neuromuscular fatigue in autoimmune myasthenia gravis: a cross-sectional study[J]. Neurophysiologie Clinique, 2023, 53 (4): 102844 doi: 10.1016/j.neucli.2023.102844

[10]	LIU Q, LIU Y, ZHANG C, et al sEMG-based dynamic muscle fatigue classification using SVM with improved whale optimization algorithm[J]. IEEE Internet of Things Journal, 2021, 8 (23): 16835- 16844 doi: 10.1109/JIOT.2021.3056126

[11]	WANG S, TANG H, WANG B, et al Analysis of fatigue in the biceps brachii by using rapid refined composite multiscale sample entropy[J]. Biomedical Signal Processing and Control, 2021, 67: 102510 doi: 10.1016/j.bspc.2021.102510

[12]	WEI C, WANG H, ZHOU B, et al sEMG signal-based lower limb movements recognition using tunable Q-factor wavelet transform and kraskov entropy[J]. IRBM, 2023, 44 (4): 100773 doi: 10.1016/j.irbm.2023.100773

[13]	XI X, DING J, WANG J, et al Analysis of functional corticomuscular coupling based on multiscale transfer spectral entropy[J]. IEEE Journal of Biomedical and Health Informatics, 2022, 26 (10): 5085- 5096 doi: 10.1109/JBHI.2022.3193984

[14]	MENGARELLI A, TIGRINI A, FIORETTI S, et al On the use of fuzzy and permutation entropy in hand gesture characterization from EMG signals: parameters selection and comparison[J]. Applied Sciences, 2020, 10 (20): 7144 doi: 10.3390/app10207144

[15]	XIE H B, GUO J Y, ZHENG Y P Fuzzy approximate entropy analysis of chaotic and natural complex systems: detecting muscle fatigue using electromyography signals[J]. Annals of Biomedical Engineering, 2010, 38 (4): 1483- 1496 doi: 10.1007/s10439-010-9933-5

[16]	侯言旭, 姜礼杰, 胡保华, 等基于边际谱熵的肌肉疲劳实时评估方法研究[J]. 仪器仪表学报, 2017, 38 (7): 1625- 1633 HOU Yanxu, JIANG Lijie, HU Baohua, et al Real-time assessment of muscle fatigue based on marginal spectrum entropy[J]. Chinese Journal of Scientific Instrument, 2017, 38 (7): 1625- 1633

[17]	石欣, 余可祺, 敖钰民, 等基于下肢sEMG的疲劳模糊增量熵表征方法研究[J]. 仪器仪表学报, 2024, 45 (5): 271- 280 SHI Xin, YU Keqi, AO Yumin, et al Research on entropy of incremental fuzzy entropy representation model for lower limb fatigue based on sEMG[J]. Chinese Journal of Scientific Instrument, 2024, 45 (5): 271- 280 doi: 10.19650/j.cnki.cjsi.J2412468

[18]	BERETTA-PICCOLI M, BOCCIA G, PONTI T, et al Relationship between isometric muscle force and fractal dimension of surface electromyogram[J]. BioMed Research International, 2018, 2018: 5373846 doi: 10.1155/2018/5373846

[19]	JIANG W, XIA C, ZHANG Y, et al. Research on muscle fatigue trend via nonlinear dynamic feature analysis of mechanomyography signal [C]//Proceedings of the IEEE 4th International Conference on Signal and Image Processing. Wuxi: IEEE, 2019: 669–673.

[20]	SU Z, LIU H, QIAN J, et al Hand gesture recognition based on sEMG signal and convolutional neural network[J]. International Journal of Pattern Recognition and Artificial Intelligence, 2021, 35 (11): 2151012 doi: 10.1142/S0218001421510125

[21]	WANG G, JIN L, ZHANG J, et al Recurrent neural network enabled continuous motion estimation of lower limb joints from incomplete sEMG signals[J]. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 2024, 32: 3577- 3589 doi: 10.1109/TNSRE.2024.3459924

[22]	SHARMA T, SHARMA K P Novel approach of time series prediction using SEMG on hand gestures for prosthetic control[J]. Sādhanā, 2025, 50 (3): 121 doi: 10.1007/s12046-025-02759-1

[23]	MU D, LI F, YU L, et al Study on exercise muscle fatigue based on sEMG and ECG data fusion and temporal convolutional network[J]. PLoS One, 2022, 17 (12): e0276921 doi: 10.1371/journal.pone.0276921

[24]	HWANG S, KWON N, LEE D, et al A multimodal fatigue detection system using sEMG and IMU signals with a hybrid CNN-LSTM-attention model[J]. Sensors, 2025, 25 (11): 3309 doi: 10.3390/s25113309

[25]	WANG J, SUN S, SUN Y A muscle fatigue classification model based on LSTM and improved wavelet packet threshold[J]. Sensors, 2021, 21 (19): 6369 doi: 10.3390/s21196369

[26]	LANGEVIN A, CARBONNEAU M A, CHERIET M, et al Energy disaggregation using variational autoencoders[J]. Energy and Buildings, 2022, 254: 111623 doi: 10.1016/j.enbuild.2021.111623

[27]	AZAMI H, ESCUDERO J Amplitude- and fluctuation-based dispersion entropy[J]. Entropy, 2018, 20 (3): 210 doi: 10.3390/e20030210

[28]	JARRETT D, YOON J, VAN DER SCHAAR M Dynamic prediction in clinical survival analysis using temporal convolutional networks[J]. IEEE Journal of Biomedical and Health Informatics, 2019, 24 (2): 424- 436

[29]	AL-SELWI S M, HASSAN M F, ABDULKADIR S J, et al RNN-LSTM: from applications to modeling techniques and beyond: systematic review[J]. Journal of King Saud University: Computer and Information Sciences, 2024, 36 (5): 102068 doi: 10.1016/j.jksuci.2024.102068

[30]	曹震, 吕东澔, 张勇, 等基于sEMG信号几何特征的肌肉疲劳分类[J]. 传感器与微系统, 2024, 43 (7): 145- 148 CAO Zhen, LV Donghao, ZHANG Yong, et al Muscle fatigue classification based on geometric features of sEMG signal[J]. Transducer and Microsystem Technologies, 2024, 43 (7): 145- 148 doi: 10.13873/J.1000-9787(2024)07-0145-04

[31]	徐哲熙, 刘婷, 任晟民, 等基于时移多尺度波动散布熵和改进核极限学习机的水电机组故障诊断[J]. 工程科学与技术, 2024, 56 (3): 41- 51 XU Zhexi, LIU Ting, REN Shengmin, et al Fault diagnosis method of hydropower units based on time-shifted multiscale fluctuation dispersion en-tropy and improved kernel extreme learning machine[J]. Advanced Engineering Sciences, 2024, 56 (3): 41- 51 doi: 10.12454/j.jsuese.202200843

[1]	Naizhou ZHANG,Yunchao ZHAO,Wei CAO,Xiaojian ZHANG. Image captioning generation based on multiple-view cross-modal feature fusion[J]. Journal of ZheJiang University (Engineering Science), 2026, 60(6): 1205-1212.

[2]	Wen XUE,Shuo ZHAO,Yongqiang LI. Imitation learning for bipedal robots based on simplified probabilistic framework for options[J]. Journal of ZheJiang University (Engineering Science), 2026, 60(5): 1071-1081.

[3]	Guoyan LI,Wei YU,Yupeng MEI,Minghui ZHANG,Xinqiang WANG. Building extraction from remote sensing images with global-local feature fusion[J]. Journal of ZheJiang University (Engineering Science), 2026, 60(5): 1100-1108.

[4]	Tianhe YU,Wenlong WANG,Yong LIU,Zhuangzhuang YANG,Shanchong HOU. Improved algorithm for identifying occluded vehicles and pedestrians in foggy images[J]. Journal of ZheJiang University (Engineering Science), 2026, 60(4): 738-750.

[5]	Long MA,Yongqi HOU,Baijing WU,Li GAO,Jianwei DENG,Guanghui YAN. Water-floating garbage trajectory prediction model based on multi-scale graph convolution[J]. Journal of ZheJiang University (Engineering Science), 2026, 60(4): 751-762.

[6]	Xiaoan BAO,Enlin CHEN,Na ZHANG,Xiaomei TU,Biao WU,Qingqi ZHANG. 3D human pose estimation based on multi-scale encoder fusion[J]. Journal of ZheJiang University (Engineering Science), 2026, 60(3): 565-573.

[7]	Guoyan LI,Penghui LI,Rong LIU,Yupeng MEI,Minghui ZHANG. Remote sensing road extraction by fusing multi-scale resolution and strip feature[J]. Journal of ZheJiang University (Engineering Science), 2026, 60(3): 585-593.

[8]	Jian XIAO,Xinze HE,Hongliang CHENG,Xiaoyuan YANG,Xin HU. Aerial small target detection algorithm based on multi-scale feature enhancement[J]. Journal of ZheJiang University (Engineering Science), 2026, 60(1): 19-31.

[9]	Siyao ZHOU,Nan XIA,Jiahong JIANG. Pose-guided dual-branch network for clothing-changing person re-identification[J]. Journal of ZheJiang University (Engineering Science), 2026, 60(1): 71-80.

[10]	Chaoqun DONG,Zhan WANG,Ping LIAO,Shuai XIE,Yujie RONG,Jingsong ZHOU. Lightweight YOLOv5s-OCG rail sleeper crack detection algorithm[J]. Journal of ZheJiang University (Engineering Science), 2025, 59(9): 1838-1845.

[11]	Wenxin CHENG,Guanghui YAN,Wenwen CHANG,Baijing WU,Yaning HUANG. Channel-weighted multimodal feature fusion for EEG-based fatigue driving detection[J]. Journal of ZheJiang University (Engineering Science), 2025, 59(9): 1775-1783.

[12]	Zhuguo ZHOU,Yujun LU,Liye LV. Improved YOLOv5s-based algorithm for printed circuit board defect detection[J]. Journal of ZheJiang University (Engineering Science), 2025, 59(8): 1608-1616.

[13]	Jiarui FU,Zhaofei LI,Hao ZHOU,Wei HUANG. Camouflaged object detection based on Convnextv2 and texture-edge guidance[J]. Journal of ZheJiang University (Engineering Science), 2025, 59(8): 1718-1726.

[14]	Aiying HUANG,Xiaohui LI,Shuxian SUN,Yiqun ZHU. Causal disentanglement-based domain adaptation for rolling bearing fault diagnosis[J]. Journal of ZheJiang University (Engineering Science), 2025, 59(7): 1523-1531.

[15]	Shenchong LI,Xinhua ZENG,Chuanqu LIN. Multi-task environment perception algorithm for autonomous driving based on axial attention[J]. Journal of ZheJiang University (Engineering Science), 2025, 59(4): 769-777.

Viewed

Full text

Abstract

Cited

Shared

Discussed