中国组织工程研究 ›› 2025, Vol. 29 ›› Issue (30): 6489-6498.doi: 10.12307/2025.924

• 组织构建与生物力学 tissue construction and biomechanics • 上一篇    下一篇

基于鱼群算法对运动者疲劳步态的动作识别

张  健1,蔡  峰2,李婷文1,任鹏博3   

  1. 1闽南师范大学体育学院,福建省漳州市  363000;2广东科技学院体育部,广东省东莞市  523083;3常州大学怀德学院,江苏省常州市  213164
  • 收稿日期:2024-10-16 接受日期:2025-11-26 出版日期:2025-10-28 发布日期:2025-03-28
  • 通讯作者: 任鹏博,博士,讲师,常州大学怀德学院,江苏省常州市 213164
  • 作者简介:张健,男,1984年生,2020年华东师范大学毕业,博士,副教授,硕士生导师,福建省高层次人才,“龙江计划”优青人才,主要从事人体运动与健康促进方面的研究。
  • 基金资助:
    教育部人文社会科学研究基金项目(23YJCZH293),项目负责人:张健;福建省社会科学研究基金项目(FJ2022B021),项目负责人:张健

Fatigue gait recognition of athletes based on fish swarm algorithm

Zhang Jian1, Cai Feng2, Li Tingwen1, Ren Pengbo3   

  1. 1School of Physical Education, Minnan Normal University, Zhangzhou 363000, Fujian Province, China; 2Department of Sports, Guangdong University of Science and Technology, Dongguan 523083, Guangdong Province, China; 3Huaide College, Changzhou University, Changzhou 213164, Jiangsu Province, China 
  • Received:2024-10-16 Accepted:2025-11-26 Online:2025-10-28 Published:2025-03-28
  • Contact: Ren Pengbo, PhD, Lecturer, Huaide College, Changzhou University, Changzhou 213164, Jiangsu Province, China
  • About author:Zhang Jian, PhD, Associate professor, Master’s supervisor, School of Physical Education, Minnan Normal University, Zhangzhou 363000, Fujian Province, China
  • Supported by:
    Humanities and Social Sciences Research Fund Project of Ministry of Education, No. 23YJCZH293 (to ZJ); Fujian Social Science Research Fund Project, No. FJ2022B021 (to ZJ)

摘要:



文题释义:
步态能量图:在步态分析中,步态能量图是一种有效的特征表示方法,其通过对一个完整步态周期内的图像帧进行累加平均,量化步态周期内身体各部位的运动能量分布,不仅直观展示了运动者在不同运动阶段的步态特征,还为运动者的疲劳状态识别和训练调整提供了科学依据。通过计算步态周期内各像素点的能量值,可以生成一张包含丰富步态信息的图像,便于后续的分析和识别。
鱼群算法:是一种基于自然界鱼群行为的启发式优化算法,用于求解复杂问题的最优解或近似最优解。在运动者步态动作识别模型参数求解中,鱼群算法通过模拟鱼群的觅食、聚群、追尾和随机等行为,在解空间中探索最优参数配置。每条“鱼”代表一个潜在的参数解,通过不断迭代和更新,最终找到全局最优解。鱼群算法具有良好的全局搜索能力和并行搜索效率,能够突破局部最优解,为步态识别模型提供最佳的初始参数配置。

背景:步态动作是运动中展现出的重要特征之一,反映了身体状态和运动能力。在疲劳状态下步态动作会出现异常,如步幅减小、身体摇晃等,这些异常步态动作会对身体造成伤害。
目的:旨在推动运动科学领域的技术进步,将先进的算法与数据分析技术应用于运动实践中,从而进一步提升运动疲劳状态下步态动作的识别准确度。
方法:基于鱼群算法的运动疲劳状态下步态动作识别方法。利用归一化自相关函数和运动能量分布原理获取运动者单周期步态能量图,采用奇异值分解法转换图像以突出视觉差异,生成运动者步态能量图,使用卷积神经网络构建步态动作识别模型,并通过鱼群算法求解模型的参数,以提升疲劳步态动作识别的准确度及效率。
结果与结论:①鱼群算法在步态动作识别上的损失值较小,能够准确、快速地识别出运动者的步态动作,动态监测运动者身体疲劳情况;②基于鱼群算法的运动者疲劳步态动作识别研究,可以有效识别出运动者疲劳状态下的步态动作,实现对于细微步态变化的精确捕捉;③鱼群算法的系统稳定性良好,能够减少试验测试结果的波动性,提高识别效率,更有效地管理运动疲劳、预防运动损伤。另外,当正常人步态特征发生显著变化时,系统可以发出预警,提示个体可能处于疲劳状态,需要休息或调整活动强度。
https://orcid.org/0000-0001-9247-3929(张健)

中国组织工程研究杂志出版内容重点:组织构建;骨细胞;软骨细胞;细胞培养;成纤维细胞;血管内皮细胞;骨质疏松;组织工程

关键词: 鱼群算法, 运动者, 卷积神经网络, 疲劳步态, 步态能量图, 卷积核, 运动能量分布, 归一化自相关函数, 工程化组织构建

Abstract: BACKGROUND: Gait movements are one of the important characteristics exhibited by athletes during exercise, reflecting their physical condition and athletic ability. In a state of fatigue, athletes may exhibit abnormal gait movements, such as reduced stride and body shaking, which can cause harm to their bodies.
OBJECTIVE: To promote technological progress in the field of sports science by applying advanced algorithms and data analysis techniques to the training practice of athletes, so as to further improve the recognition accuracy of gait movements under sports fatigue. 
METHODS: A gait recognition method for athletes in fatigue state was based on fish swarm algorithm. By utilizing the normalized autocorrelation function and the principle of motion energy distribution, a single cycle gait energy map of athletes was obtained. Singular value decomposition was used to transform the image to highlight visual differences, generating a gait energy map of athletes. A convolutional neural network was used to construct a gait action recognition model, and the parameters of the model were solved using the fish swarm algorithm to improve the accuracy and efficiency of fatigue gait action recognition.
RESULTS AND CONCLUSION: (1) The fish swarm algorithm had a small loss value in gait action recognition, and could accurately and quickly identify the gait actions of athletes, and dynamically monitor their physical fatigue. (2) The research on fatigue gait recognition of athletes based on fish swarm algorithm could effectively identify the gait movements of athletes in fatigue state and achieve accurate capture of subtle gait changes. (3) The system stability of this method is good, which can reduce the volatility of experimental test results and improve recognition efficiency, can more effectively manage sports fatigue and prevent sports injuries. In addition, when the gait characteristics of normal people change significantly, the system can give an early warning, indicating that the individual may be in a state of fatigue and need to rest or adjust the intensity of activity.

中国组织工程研究杂志出版内容重点:组织构建;骨细胞;软骨细胞;细胞培养;成纤维细胞;血管内皮细胞;骨质疏松;组织工程

Key words: fish swarm algorithm, athletes, convolutional neural network, fatigue gait, gait energy diagram, convolutional kernel, kinetic energy distribution, normalized autocorrelation function, engineered tissue construction

中图分类号: