ABSTRACT
The instrument and the experimental environment influence the infrared spectra, which may limited the identification of the samples by a prediction model. Based on the Fourier transform infrared spectroscopy (FTIR) technology, the authors performed different infrared spectral calibration methods for Radix Zanthoxyli geographical origins determination, the SIMCA was used to establish an identification models, and the model was used to distinguish samples from four different regions of Guangxi. According to the result of prediction, the authors could obtain the most suitable calibration method for the identification model. The results showed that, respectively, by the multiple scattering correction and standard normal variation, their PCA data distribution and the distance between models is ideal, suggesting that we can eliminate the interference from the environmental and human factors by these two correction methods, and also separate each samples of different habitats. The test using the method to measure the geographical origins of Radix Zanthoxyli proved that the recognition rate and rejection rate are both at or near 100%. Visible, and both the multiplicative scatter correction and the standard normal variation are all the ideal calibration methods for Radix Zanthoxyli infrared spectral geographical origins determination.
Subject(s)
Ranunculaceae , Calibration , Geography , Plant Roots , Spectroscopy, Fourier Transform InfraredABSTRACT
In the present work, the authors explored a rapid method of the Zanthoxylum nitidum geographical origins determination. Based on Fourier transform infrared spectroscopy (FTIR) technology, the band of 1 800-400 cm(-1) which is the IR fingerprint of Zanthoxylum nitidum, the Fisher ratio and the soft independent modeling of class analogies (SIMCA) were used to build a classification model. Respectively, four kinds of Zanthoxylum nitidum in the Guangxi region were detected by the model, and the model was verified by calculating their recognition rate and rejection rate. The results show that the authors can accurately extract the overall information of Chinese herbal medicines by using the FTIR, also established a pattern recognition model to predict unknown samples, and obtained satisfactory recognition rate and rejection rate, indicating that the model has stronger ability of identification. The detection on real time was carried out rapidly with the Fisher model, suggesting that the model has more practical value.
