计算机工程与应用 ›› 2018, Vol. 54 ›› Issue (23): 120-124.DOI: 10.3778/j.issn.1002-8331.1806-0390

• 模式识别与人工智能 • 上一篇    下一篇

基于SPSO与ISKPCA的RdR散点图识别分类研究

岳大超,甘良志,刘海宽,余南南   

  1. 江苏师范大学 电气工程及自动化学院,江苏 徐州 221116
  • 出版日期:2018-12-01 发布日期:2018-11-30

Research on classification of RdR scatter plot based on SPSO and ISKPCA

YUE Dachao, GAN Liangzhi, LIU Haikuan, YU Nannan   

  1. Institute of Electrical Engineering and Automation, Jiangsu Normal University, Xuzhou, Jiangsu 221116, China
  • Online:2018-12-01 Published:2018-11-30

摘要: 面对稀缺的医疗资源,心血管疾病的上升趋势,自动化诊断日趋迫切。为实现心电自动化诊断,提出了一种使用简化粒子群算法来自动搜寻集成稀疏核主分量分析的参数,并以此获得的集成稀疏核主分量分析模型来对用心电数据绘制的RdR散点图进行识别分类的方法,以期实现心电自动化诊断。算法通过计算样本数据与使用核主分量分析映射数据之间的距离差值来研究数据之间的最大相似性,并以此来判断样本数据类别,在对正常窦性心律、非偶联早搏、二联律早搏、三联律早搏以及混合早搏这五种心律进行的分类实验结果显示,可以准确识别不同的心律,分类的正确率较高,有助于心电自动化诊断的实现。

关键词: 心电信号, 稀疏核主分量分析, RdR散点图, 智能诊断, 简化粒子群算法

Abstract: Faced with scarce medical resources and the rising trend of cardiovascular diseases, automated diagnosis is increasingly urgent. The method of identifying and classifying RdR scatter plots drawn with ECG data using integration sparse kernel principal component analysis and simplified particle swarm optimization is presented. The algorithm uses minimum error to study the maximum correlation between data, and then the ECG may be classified as NSR, or uncoupling premature beat, or two combined premature beat, or triple combined premature beat, or mixed premature beat. Experiments show that its performance is promising. This provides the basis for automatic diagnosis of ECG.

Key words: ECG, sparse kernel principal component analysis, RdR scatter plots, intelligent diagnosis, simplified particle swarm optimization