二尖瓣心音信号降噪与识别的理论仿真

Theoretical simulation on noise reduction and recognition of mitral heart sound signal

为提高心音信号在临床医学上的识别准确度,提出双谱减法和粒子蚁群聚类SVM算法。采用双谱减法对二尖瓣心音信号通过滤波、傅里叶变换以及减法操作,得到降噪后的频域信号,将频域信号进行逆傅里叶变换得到时域信号,提取其信号的特征值,完成降噪处理。采用粒子群算法找出特征数据的局部与全局极值,确定初始聚类中心,通过蚁群聚类算法优化SVM,完成心音数据的识别处理。实验结果表明,粒子蚁群聚类算法对二尖瓣心音信号的识别准确率较高。

To improve the recognition accuracy of heart sound signals in clinical medicine,a two-spectrum subtraction and particle ant colony clustering SVM algorithm was proposed.the two-point subtraction method was used to filter the mitral heart sound signal through filtering,Fourier transform and subtraction operation were implemented to obtain the frequency-domain signal after noise reduction.The frequency domain signal was inversely Fourier transformed to obtain the time domain signal,and the signal was extracted.The characteristic value of the noise reduction process was completed.The particle swarm optimization algorithm was used to find the local and global extremum of the feature data,the initial cluster center was determined,the SVM was optimized using the ant colony clustering algorithm,and the recognition process of the heart sound data was completed.Experimental results show that the particle ant colony clustering algorithm has higher recognition accuracy for mitral heart sound signals.