基于改进邻域粗糙集的PPG信号身份识别方法  

Method for PPG signal identification based on improved neighborhood rough set

作　　者：孙斌 薛毓楠 陈小惠 Sun Bin;Xue Yu'nan;Chen Xiaohui(School of Automation and Artificial Intelligence,Nanjing University of Posts and Telecommunications,Nanjing 210000,China)

机构地区：[1]南京邮电大学自动化学院、人工智能学院,南京210000

出　　处：《国外电子测量技术》2022年第6期8-13,共6页Foreign Electronic Measurement Technology

基　　金：江苏省研究生科研与实践创新计划(KYCX20_0799)项目资助。

摘　　要：针对现有生理信号身份识别方法的算法复杂性高、识别率低等问题,提出了一种基于光电容积脉搏波(photo plethysmo-graphy,PPG)信号的身份识别方法。首先对PPG信号提取了心指数、幅度差等24维特征;然后利用柯西扰动量子粒子群优化邻域粗糙集(CQPNR)算法进行特征约简与寻优,获取最佳特征子集;最后利用人工蜂群算法优化支持向量机(ABC-SVM)算法对样本的最佳特征子集进行训练与测试,完成个体身份识别。仿真结果表明,该方法的识别准确率可以达到98.9%。Aimingat the problems of high algorithm complexityand low recognition rate of existing physiological signal identification methods,an identification method based on photo plethysmo signal is proposed Firstly 24 dimensional features such as cardiac indexand amplitude difference are extracted from PPG signal.Then,the Cauchy perturbation quantum particle swarm optimization neighborhood rough set(CQPNR)algorithm is used for feature reductio nand optimization toobtain the best feature subset.Finally,theartificial beecolony algorithm is used to optimize the support vector machine(ABC-SVM)algorithm to train and test the best feature subset of thesampleto complete individual identifica tion.Simulation results show that the recognition accuracy of this method can reach 98.9%.

关 键 词：PPG信号 CQPNR算法 ABC-SVM算法 识别率 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]