摘要
采用ARMA参数化双谱估计法对15例吸毒者和15例正常人的脉象信号进行了参数化双谱估计.在得到每一例脉搏波的归一化双谱幅值对角切片后,应用K-L变换得到30个样本特征向量,然后设计4-9-1的BP神经网络进行模式识别.除1例正常人被误判外,其余的受测者全部予以正确识别.研究结果表明,归一化双谱对角切片与K-L变换相结合的方法是提取中医脉象信号特征的有效方法.这种方法与BP神经网络的联合应用,对海洛因吸毒者和正常人的脉象信号具有很高的识别率.
The ARMA parameterized bispectrum estimation method is used to estimate the bispectra of the pulse signals for 15 heroin addicts and 15 healthy persons. After obtaining a diagonal slice of the normalized bispeetrum magnitude for every pulse wave, K-L transform is used to get 30 sample feature vectors. Then a 4-9-1 BP neural network is successfully designed and used to the pattern classification for sample feature vectors. All subjects characterized by the feature vectors, except for one normal person, are identified correctly. The research result shows that the method of the diagonal slice of the normalized bispectrun magnitude and K-L transform is an effective method for extracting features from Chinese pulse signals. The application of this method associated with BP neural network has a high identification rate.
出处
《重庆工学院学报》
2007年第5期5-9,共5页
Journal of Chongqing Institute of Technology
基金
重庆市自然科学基金资助项目(CSTC
2004BB5061)
关键词
脉象信号
参数化双谱估计
K—L交换
BP网络
pulse signal
paraneterized bispectrun estimation
K-L transformation
BP neural network
