| 心电信号识别的人工神经网络方法研究 |
| |
| 引用本文： | 谭明娟,马信山. 心电信号识别的人工神经网络方法研究[J]. 电子与信息学报, 1996, 0(Z1) |
| |
| 作者姓名： | 谭明娟 马信山 |
| |
| 作者单位： | 清华大学电机系，清华大学电机系 北京 100084，北京 100084 |
| |
| 基金项目： | 高等学校博士学科点专项科研基金 |
| |
| 摘 要： | 本文提出一种新的隐含层神经元非全连接的心电信号识别的BP阿络(Back-Propagation neural network)。该网络也可以看成是多个子网络的并联,用该网络作多种正常、异常心电信号识別时,利用心电信号奉身的特点,将一周期心电信号按P波、QRS波群、T波、ST段,分成n段,每个子网络只反映心电信号某一段的病变信息,多个子网络的并联结果,综合反映整个心电信号周期的信息。实验表明:与一般多层网络相比,该网络具有更高的识别能力,更快的学习速度。
|
| 关 键 词： | 心电信号 人工神经网络 BP算法 |
A STUDY OF THE NEURAL NETWORK FOR ECG PATTERN RECOGNITION |
| |
| Tan Mingjuan Jiang Jiguang Ma Xinshan. A STUDY OF THE NEURAL NETWORK FOR ECG PATTERN RECOGNITION[J]. Journal of Electronics & Information Technology, 1996, 0(Z1) |
| |
| Authors: | Tan Mingjuan Jiang Jiguang Ma Xinshan |
| |
| Abstract: | A new non-thorough-connected neural network for electrocardioqraphic(ECG) classification is presented. In the new network, one of the second-hidden-layer units is only connected with part units of the first-hidden-layer, and this connection is controlled to focus on the specific abnomality in ECG wave. On the other hand, the new network is composed of many parallel sub-BP-networks, each sub-BP-network can reflect all the specific information on P-wave or QRS complex or T-wave and ST segment. Comparing it with the usual network, we have found that the new network has higher ability in recognition and higher speed in learning. |
| |
| Keywords: | ECG signal Artificial neural network Back-propagation algorithm |
| 本文献已被 CNKI 等数据库收录！ |
|
