Study On Applications Of Pattern Recognition In Near Infrared Spectroscopy For Traditional Chinese Medicine

Near infrared spectroscopy is a fast and efficient analytical method. In this dissertation, the research focused on the applications of near infrared spectroscopy in qualitative analysis of Traditional Chinese Medicine. Pattern recognition methods were used for processing the spectral data. The research content is as follows.1) A method based on Mahalanobis Distance using near infrared spectroscopy was proposed. Distinction of different planting areas of Radix Salviae Miltiorrhiza was investigated as an example to test the performance of this method. The planting area of each individual raw material was identified accurately. The proposed method can be used to identify the quality of complex materials.2) Combined the Support Vector Machine with the fuzzy membership of training data, a new pattern recognition method was constructed. A group of Shenmai injection spectral was investigated as an example to test the performance of this method. The classification result of proposed method is better than three other pattern recognition methods.3) A fast method based on near infrared spectroscopy was developed for on-line endpoint determination in blending process of dripping solution. The standard deviations between spectral were calculated as the index of mixture uniformity. The result shows that this method for on-line endpoint determination of dripping solution is fast and accurate.