基于支持向量机的亚健康状态识别  

Recognition of Sub-health Based on Support Vector Machine

作　　者：杨凤霞[1] 

机构地区：[1]西安航空职业技术学院,陕西西安710089

出　　处：《现代电子技术》2009年第20期167-170,共4页Modern Electronics Technique

摘　　要：利用HK-2000C集成化数字脉搏传感器提取人体左关处桡动脉脉搏信号,然后计算脉搏功率谱,并在此基础上提取功率谱峰值、功率谱重心及其对应频率的特征量,最后利用支持向量机(SVM)对所提特征进行分类。通过与线性判别式分析(LDA)法分类结果对比可以看出,SVM算法有着理论上的突出优势,但在实际应用中,由于样本数目有限,而且在其核函数选择和参数调整方面,均需要视经验值而定,因此推广性较差,还需要进一步的研究和改进。Pulse signal of radial artery are picked up by using HK - 2000C digital integrated pulse transducer. And power spectrum is calculated. Then peak value, peak frequency, center of gravity （cg） and gravity frequency of power spectrum are extracted. And Support Vector Machine （SVM） is applied to pattern recognition of sub- health. Finally, the results of Linear Discriminant Analysis （LDA） and SVM are compared. Though there are advantages in theory using SVM. Some problems in selection of kernel parameter which usually selected by experience. There still has more works should be done.

关 键 词：亚健康 脉象 功率谱 支持向量机 

分 类 号：TP274.3[自动化与计算机技术—检测技术与自动化装置]