基于改进遗传模拟退火K-means的心电波形的分类研究

Research of ECG waveforms classification based on improved genetic simulated annealing K-means

针对心电图自动诊断困难这一问题,提出了一种新的聚类算法:基于均方差属性加权的遗传模拟退火K-means改进聚类算法,用于改进心电图(ECG)信号的自动识别技术。利用小波变换的多分辨率和抗干扰能力好的特点,检测QRS波、P波、T波,提高了特征检测的准确性;利用聚类分析具有较好的鲁棒性和适合于大数据量分析的特点,对心电信号进行波形分类。采用MIT-BIH标准心电数据库中的部分数据对识别结果进行判断,改进后的K-means聚类算法的准确率高于传统的K-means聚类算法,实验表明该算法对心电信号可以进行有效分类。

In view of the difficulties to recognize ECG signal automatically,this paper presented a new clustering algorithm, which was proposed based on the MSE attribute weights genetic simulated annealing to improve K-means clustering algorithm , in order to improve the ECG signal automatic identification technology.It used wavelet transform and multi-resolution and good anti-jamming capability to detect QRS complex,P wave,T wave,improved the accuracy of feature detection.Because of the cluster method had more robust and suitable for large data volume analysis,it classified the ECG signals by using this method to analyze large data volume.It adopted the parts of data from the MIT-BIH standard ECG database to judge the result of the i-dentification.The improved K-means clustering algorithm is more accurate than the traditional K-means clustering algorithm, experiments indicate that this algorithm is effective and accurate to classify ECG signals.