基于希尔伯特-黄变换和提升小波包相结合的心音信号识别研究

目的：研究基于希尔伯特-黄变换和提升小波包相结合的方法对正常和异常心音信号识别的效果。方法：首先用DB6小波对心音信号进行降噪处理,然后用希尔伯特-黄变换分析提取心音信号的时域、频域特征值,再通过自适应提升小波包提取信号的频带能量特征值,最后通过支持向量机对心音信号进行分类识别。结果：对临床采集的240例异常心音和正常心音进行实验,正确识别率达到97.2%,且运算速度很快。结论：希尔伯特-黄变换和自适应提升小波包相结合的方法可有效识别正常和各种异常的心音信号,值得推广应用。

Study on heart sound identification based on combined Hilbert-Huang transform and lifting wavelet package

Objective To improve recognition rate of normal and abnormal heart sound signal by the method combing Hilbert-Huang transform and lifting wavelet package. Methods DB6 wavelet was used to denoise the heart sound signal firstly, then HHT transform was applied to extracting the characteristic values for frequency domain and time domain,and self-adaptive lifting wavelet package was mobilized to acquire the characteristic values of the frequency band energy, finally support vector machine was utilized for the classified recognition of the signals. Results Totally 240 cases of abnormal and normal heart sound signals underwent the experiment, with the recognition rate of 97.2% and high calculation speed. Conclusion HHT combined with lifting wavelet package can recognize kinds of normal and abnormal heart sound signals, and thus is worth popularizing practically.