J4 ›› 2013, Vol. 31 ›› Issue (1): 38-44.

• 论文 • 上一篇    下一篇

下肢康复器械SVM建模的关键技术

刘通1,2, 李海富2, 王丽荣2, 臧睦君3   

  1. 1. 长春理工大学 电子信息工程学院, 长春 130022;2. 长春大学 电子工程学院, 长春 130022;3. 吉林大学 通信工程学院, 长春 130012
  • 收稿日期:2012-09-17 出版日期:2003-01-24 发布日期:2013-04-01
  • 作者简介:刘通(1987—), 男, 吉林双辽人, 长春理工大学硕士研究生, 主要从事模式识别与智能系统研究, (Tel)86-15144151446(E-mail)q285441338@126.com|通讯作者[HT6SS]: 李海富(1964—), 男, 长春人, 长春大学副教授, 主要从事智能控制与医疗器械研究, (Tel)86-18604319639(E-mail)haifuli@163.com。
  • 基金资助:

    吉林省教育厅基金资助项目(吉教科合字[2012]第243号)

Key Technology of Lower Limb Rehabilitation SVM Modeling

LIU Tong1,2, LI Hai-fu2, WANG Li-rong2, ZANG Mu-jun3   

  1. 1. College of Electronic and Information Engineering, Changchun University of Science and Technology,Changchun 130022, China|2. School of Electronic Engineering, Changchun University, Changchun 130022, China;3. College of Communication Engineering, Jilin University, Changchun 130012, China
  • Received:2012-09-17 Online:2003-01-24 Published:2013-04-01

摘要:

为解决支持向量机用于下肢康复训练器驱动系统建模的训练参数选取问题, 使用Simulink平台采集样本, 通过Libsvm工具箱的相关接口进行对比实验, 选取回归问题中较先进的3种目标函数寻优算法, 讨论算法在建模中的应用范围和价值, 并提出选取依据。研究结果表明, 网格搜索和粒子群算法均方误差均接近于零, 相关系数大于99%, 适用于需求稳定和实用性强的系统; 遗传算法不稳定, 但在实验室仿真阶段研究支持向量机性能中亦有其实用性。该方法有效地解决了参数选取问题, 同时也为进一步研究面向对象的目标函数极值寻优算法改进提供了基础。

关键词: 系统建模, 康复训练器, 支持向量机, 训练参数

Abstract:

The support vector machine is used in lower limb rehabilitation training for device driving system modeling, one of the important issues is to select training parameters. In order to solve this issue, using the simulink platform collected samples, and through the libsvm toolbox, the related interface is experimented. In contrast experiments, three kinds of regression of advanced objective function optimization algorithm are selected. Algorithm in modeling application scope and value is discussed, and the basis for selecting is given. The grid search and particle swarm optimization have a mean square error close to zero, the correlation coefficient is greater than 99%. It applies to the demand of stability and practicality of the system. The genetic algorithm is not stable, but also has practical application. The results show that the method effectively solves the problem of selecting parameters, and at the same time for the further study of the object oriented objective function external optimization algorithm improvements
, especially for high requirements of real-time dynamic identification condition}, provide some basic support.

Key words: system modeling, rehabilitation training device, support vector machine, training parameters

中图分类号: 

  • TP181