Applied Research Of Pattern Recognition In Chinese Traditional Medicine Fingerprint Identification And The Environmental Data

This thesis was about the application of chemical pattern recognition to acquirement of Chinese traditional medicine fingerprint identification, to estimate of default in environmental data and to searching the environmental polluted sources.First, the theory on pattern recognition was introduced, and then the application of pattern recognition in chemistry, especially in Chinese traditional medicine analysis and environmental chemistry, was reviewed in detail.Chinese traditional medicine fingerprint identification guarantees the quality control of Chinese traditional medicine, its modernization and being world-class. It is such a spectrum that is obtained by spectroscopy analysis or chromatogram in order to generally disclose the sort and amount of components in Chinese traditional medicine and then to show its quality. Our work was to select fingerprint peaks from capillary gas chromatography data of Yuxingcao injection by clustering analysis and then to construct the fingerprint identification. The test results showed that the fingerprint identification could identify Yuxingcao correctly.Another part of our work was to estimate default in environmental data and to find environmental polluted sources by pattern recognition. Pattern recognition technique is a kind of non-parameter prediction method that is more fitted for the type of data that is not satisfied with normal distribution. We estimated the default in the river water normal detection data by KNN (kith nearest neighbor) in pattern recognition. Then we made a comparison with other methods. Our method was to let every sample be discrete point in N-dimension space. The same kind of data would cluster in this space. If there were k nearest neighbor points, we could obtain the default by calculating the all distances between them. Finally we found environmental polluted sources by PCR and factor analysis. The good results were gotten through rotating abstract factors to actual factors and analyzing the environment polluted sources and trends from the environmental data. We found that there were different factors in different areas after analyzing the river water. So the pertinent management could be implemented to different areas.