摘要
目的:针对已有心电数据压缩方法开销较大、难以用于工程实践等问题,提出一种基于BP神经网络的心电数据压缩方法。方法:基于BP神经网络的思想,建立两个3层漏砂型前馈人工神经元网络,将单个心搏分为P波、QRS波和T波,分别对3个波用两个人工神经网络系统进行压缩,采用不完全联结结构提高神经网络压缩算法的波形重现能力和抗干扰能力。结果:利用BP神经网络对心电信号数据进行压缩可以实现较高的压缩比,能有效提高波形的重现能力和抗干扰的能力。结论:该方法可以有效地对心电信号采集器采集到的心电信号进行滤波和压缩等预处理,能较好地用于工程实践。
Objective Aimed at the problems that the costs of existing ECG data compression methods is high, and they are difficult to apply in engineering practice, a sort BP neural network is set up based on ECG data compression method. Methods Based on BP network theory, two three-layered feedforward neural networks were set up. Then every one heartbeat was divided into three waves, that is, P, QRS and T ones, and the three waves were compressed by two three-layered feedforward neural network individually. In order to improve the replay capability and interference rejection capability of the neural network compress algorithm, incompletely connected structure is employed. Results The method could realize high compress ratio, and improve the replay capability and interference rejection capability of the heartbeat waves. Conclusion Upon with the heartbeat signals, the method can filter and compress waves effectively, and can be used in engineering practice as well.
出处
《医疗卫生装备》
CAS
2008年第9期19-21,共3页
Chinese Medical Equipment Journal
基金
贵州省科学技术基金项目(黔科合J字[2232]号)
关键词
心电信号
数据压缩
神经网络
ECG signal
data compression
neural network
