摘要
应用变学习速率、变动量因子的BP神经网络和传统小波变换相结合,构造了一种小波神经网络,实现了高压缩比的ECG数据压缩,该算法除了具有泛化能力强、收敛速度快的特点外,还兼具了多分辨和自适应特性,有较强的特征提取能力。实验结果表明使用该算法进行ECG压缩,可以获得较高的压缩比和保真度,并且复杂度低,在异常ECG波形出现时仍可保持较好的实时性。
Combining the alterable study rate and variable momentum factor of BP neural network with traditional wavelet transformation, a new BP wavelet neural network is constructed and the higher compression ratio of ECG data is realized. It enjoys the merits of good generalization ability and high converging speed. Multi-resolution and self-adaptation are also gained, moreover, it has a strong ability of feature extraction. The experimental result indicates that this algorithm can be used to carry on the ECG compression, and obtain a desirable compression ratio and high fidelity, in addition, the complexity is low and it can still keep good real-time character while abnormal ECG wave appears.
出处
《山东科技大学学报（自然科学版）》
CAS
2007年第1期57-60,共4页
Journal of Shandong University of Science and Technology(Natural Science)
关键词
ECG
小波神经网络
压缩
重构
ECG
wavelet neural network
compressions, reconstruction
