计算机工程与应用 ›› 2007, Vol. 43 ›› Issue (20): 194-196.

• 工程与应用 • 上一篇    下一篇

基于概率神经网络的中医脉象识别方法研究

郭红霞1,王炳和1,郑思仪1,师义民2   

  1. 1.武警工程学院基础部 通信工程系,西安710086
    2.西北工业大学 应用数学系,西安710072
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2007-07-11 发布日期:2007-07-11
  • 通讯作者: 郭红霞

Recognition method of traditional Chinese medicine pulse conditions based on probabilistic neural network

GUO Hong-xia1,WANG Bing-he1,ZHENG Si-yi1,SHI Yi-ming2   

  1. 1.Engineering College of the Chinese People’s Armed Police Force,Xi’an 710086,China
    2.Applied Mathematics Department of Northwestern Polytechnical University,Xi’an 710072,China
  • Received:1900-01-01 Revised:1900-01-01 Online:2007-07-11 Published:2007-07-11
  • Contact: GUO Hong-xia

摘要: 中医脉象客观化研究的关键在于对各种脉象进行客观检测和正确识别。在长期对中医脉象进行临床检测、采集和分析的基础上,针对中医脉象模糊性强、种类繁多、特征复杂的特点,以及传统识别方法和BP神经网络识别方法的不足,提出了一种基于概率神经网络(Probabilistic Neural Network,简称PNN)的中医脉象识别方法;运用所建立的PNN脉象识别模型对中医常见的12种脉象进行了识别和检验,识别正确率平均达93%(而传统模糊聚类方法的为75%,BP神经网络方法的为87.1%)。最后对PNN方法和BP神经网络方法的识别性能做了对比实验,发现在强噪声干扰下PNN方法对脉象的识别正确率远高于BP神经网络方法。

关键词: 中医脉象, 模式识别, PNN, Bayes准则

Abstract: The key of objectivity study in Traditional Chinese Medicine(TCM) pulse-condition lies on the objective detection and correct recognition of all kinds of TCM pulse conditions.Three projects of NNSFC(National Natural Science Foundation of China)and a similar foundation of Shaanxi province have supported this research since 1988 on detection and recognition of TCM pulse conditions.In the light of the ambiguity,variety and complexity of TCM pulse conditions,and the shortcomings of the traditional recognition methods and BP neural network method,a kind of TCM pulse-condition recognition method based on Probabilistic Neural Networks(PNN) is put forward.In this research three experienced practitioners of Xi’an TCM Hospital gives us their expert advice and determined the percentage of correct recognition of our PNN method.For the twelve kinds of TCM pulse conditions,we attain an average recognition accuracy of about 93%,better than the average recognition accuracy of about 75% attained with the traditional fuzzy cluster method and 87.1% with BP Neural Network method.

Key words: traditional Chinese medicine pulse condition, pattern recognition, Probabilistic Neural Networks, Bayes’ criterion