基于SVM的触觉反馈人体姿态平衡矫正系统研究  

Research on Human Posture Balance Correction System Based on Tactile Feedback Based on SVM

作　　者：李月祥 张佳帅 焦慧敏 郭兴伟 LI Yuexiang;ZHANG Jiashuai;JIAO Huimin;GUO Xingwei(School of Mechanical and Electrical Engineering,Beijing Institute of Graphic Communication,Beijing 102600,China;School of Sports Engineering,Beijing Sport University,Beijing 100084,China)

机构地区：[1]北京印刷学院机电工程学院,北京102600 [2]北京体育大学体育工程学院,北京100084

出　　处：《北京印刷学院学报》2025年第3期67-71,共5页Journal of Beijing Institute of Graphic Communication

摘　　要：为了提高触觉反馈中人体姿态矫正的系统精度,本文采用支持向量机(SVM)算法判断人体站立状态,通过温度反馈刺激人体皮肤,实现对人体姿态的矫正。该系统首先通过自主设计的可穿戴装置实时采集人体重心偏移信息,经过分析空间各方向的多维数据,选定加速度作为特征进行前后倾姿态的识别,采用SVM算法完成平衡判定,实现触觉刺激反馈和系统决策。本文分别采用阈值法和SVM算法作为平衡判定算法进行对比实验,实验结果表明,采用阈值法前后倾平衡矫正准确率分别为93%和85.5%;采用SVM算法的前后倾斜矫正率均达到90%以上,且比阈值法情况下提高了2%和5.5%。In order to improve the systematic accuracy of human posture correction in haptic feedback,this paper uses the Support Vector Machine(SVM)algorithm to judge the standing state of the human body,and stimulates the human skin through temperature feedback to realize the correction of human posture.The system first collects the center of gravity shift information of the human body in real time through the self-designed wearable device,analyzes the multi-dimensional data in all directions of the space,selects the acceleration as the feature for the recognition of the forward and backward tilt posture,and uses the SVM algorithm to complete the balance judgment to realize tactile stimulus feedback and system decision-making.In this paper,the threshold method and SVM algorithm are used as balance judgment algorithms for comparative experiments,and the experimental results show that the accuracy of the balance correction of the threshold method is 93%and 85.5%,respectively.The front and rear tilt correction rates of SVM algorithm reached more than 90%,which were 2%and 5.5%higher than those of the threshold method.

关 键 词：触觉反馈 平衡矫正 支持向量机 阈值法 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]