摘要
为实现心音信号的自动分类,本文采用梯度提升树LightGBM对心音信号进行识别并分类。首先对心音信号进行分割,识别心脏搏动周期中的不同阶段;再分别从时域和频域对每次心跳中的不同时期片段进行分析和处理,提取信号特征,最后利用LightGBM分类器进行分类。实验结果表明,通过特征提取并利用LigthGBM进行分类是一种有效可行的心音信号分类方法,该方法对心音信号有无异常的识别准确率达91.2%。
In order to realize the automatic classification of heart sound signals,the gradient boosting tree LightGBM is used to identify and classify the heart sound signals.Firstly,the heart sound signal is segmented,and different states in the heart beat cycle are identified.Then,the segments of different state periods in each heart beat are analyzed and processed from the time domain and the frequency domain respectively,and features are extracted from the time domain and the frequency domain.Finally,LightGBM is used to classify the signals.The experimental results show that the accuracy of recognition of heart sound signals is 91.2%.
作者
余强
黄晓林
YU Qiang;HUANG Xiaolin(School of Electronic Science and Engineering,Nanjing University,Nanjing 210023,Jiangsu,China)
出处
《陕西师范大学学报（自然科学版）》
CAS
CSCD
北大核心
2020年第6期47-55,共9页
Journal of Shaanxi Normal University：Natural Science Edition
基金
江苏省自然科学基金(BK20191250)
江苏省重点研发计划(BE2017679)
南京市校企合作项目(201722002)。
