面向登月航天服基于SOM-K-Means聚类的人体步态识别
摘要
登月航天服能否精准识别航天员行走步态,并在下肢受冲击瞬间给予缓冲,成为保障航天员安全作业的关键要素.本文针对此问题展开研究,对人体步态相位进行划分,并在模拟月表低重力环境下采集步态数据.运用 Z-score 标准化和高斯滤波算法进行数据预处理.采用自组织映射(self-organizing map,SOM)-K-means 聚类算法进行步态识别模型的建模,并对模型进行评估.实验结果及聚类算法对比研究验证了 SOM-K-means 聚类算法对步态识别的有效性及模型的泛化性.结果表明,该模型的聚类效果好,步态识别准确,为高性能登月航天服的设计提供了技术支持.
Whether the lunar landing spacesuit can accurately recognize the astronaut's gait and provide timely buffering upon lower limb impact is a critical factor in ensuring the safety of extravehicular operations.This study addresses this issue by segmenting the human gait phases and collecting gait data in a simulated lunar surface low-gravity environment.Z-score normalization and Gaussian filtering are applied for data preprocessing.The self-organizing map(SOM)-K-means clustering algorithm is adopted to establish the gait recognition model,and the model is then evaluated.Experimental results and comparisons with other clustering algorithms validate the effectiveness of the SOM-K-means algorithm for gait recognition and the generalizability of the model.The results demonstrate that the model achieves satisfactory clustering performance and accurate gait recognition,providing technical support for the design of high-performance lunar landing spacesuits.
年,卷(期):
2026, 43(4)
分类号:
TP391
V445.3
机标分类号:
TP391.41
TP242.62
TP183
在线出版日期:
2026年9月20日