摘要
为更好地将表面肌电信号应用于智能轮椅的人机接口,提出了一种基于SVM的表面肌电信号动作模式的识别算法。采用一对一的方式构造SVM多值分类器,按照投票原则确定测试样本的类别归属,并与动作模式识别的核fisher算法和RBF神经网络算法进行了对比分析。实验结果表明,支持向量机(SVM)算法识别率更高,可以取得理想的学习效果和泛化性能,很好地解决小样本、非线性及局部极小值问题。
In order to apply SEMG signal to smart wheelchair machine interface, a kind of SEMG pattern recognition algorithms based on support vector machine (SVM) is put forward. Using a way of one by one to constructe SVM multi-class classifier and in accordance with the principle of voting to determine the type of test sample belongs, and with the pattern recognition of kernel fisher algorithm and RBF neural network algorithm are compared and analyzed. Experimental results show that support vector machine (SVM) algorithm recognition rate is higher, it can achieve the desired learning outcomes and generalization performance, solve the problem of small sample, nonlinear and local minima.
出处
《科学技术与工程》
北大核心
2014年第7期241-243,248,共4页
Science Technology and Engineering
关键词
支持向量机
模式分类
表面肌电信号
support vector machine (SVM) pattern classification surface electromyography