| This study mainly used Fourier transform near-infrared combined with chemometrics software,and has done the following four parts of research:1.We established the Rapid nondestructive near-infrared qualitative clustering model of four different grades of Chidi ginseng,wild ginseng,transplantation of wild ginseng and Pahuo ginseng,so that we can determine the type of ginseng on the molecular level.The recognition rates of four kinds of ginseng samples by near-infrared rapid non-destructive qualitative clustering model were 98%,98%,97% and 97%,respectively.The scores of the four types of ginseng samples on the first principal component(PC1)were 0.089724~0.12583,0.09837~0.12775,0.045894~0.18613,0.084432~0.13141,respectively.On the second principal component(PC2),the scores were-0.2904~0.16899,-0.31065~0.2116,-0.1658~0.11787,-0.20084~0.1603,and the scores on the third principal component(PC3)were-0.19041~0.32263,-0.255~0.2011,-0.35939~0.22726,-0.2113~0.28693.The classification rate of the model for the verification set of the four types of samples was 94%,indicating that the established near-infrared nondestructive rapid qualitative clustering model achieved the purpose of qualitative detection for the four types of different grades of ginseng,and then combined with the result of morphological recognition,we can be more accurately decide the level of ginseng,reducing the controversy in the process of ginseng grade evaluation.2.The transplantation of wild ginseng samples from tonghua city,jilin province and huanren manchu autonomous county,liaoning province,were analyzed for their origin traceability,and a near-infrared rapid qualitative clustering model was established for the transplanted of wild ginseng samples with different areas.The model had a 91% recognition rate on the transplantation of wild ginseng samples from tonghua city,jilin Province and the recognition rate for the transplantation of wild ginseng sample from huanren manchu autonomous county,liaoning province was 96%.The transplantation of wild ginseng samples produced in tonghua city,jilin province had scores of 0.061096~0.12355 on the first principal component(PC1)and scores of-0.119~0.096737 on the second principal component(PC2),the scores on the third principal component(PC3)were-0.25944~0.031627;The transplantation of wild ginseng samples from huanren manchu autonomous county,liaoning province had scores of-0.14058~-0.084504 on the first principal component(PC1)and scores of-0.11468~0.18757 on the second principal component(PC2),the scores on the third principal component(PC3)were-0.23505~0.16466.The classification rate of the model for the validation collection was 91%,which indicated that the established near-infrared non-destructive rapid qualitative clustering model can achieve the qualitative clustering of the transplantation of wild ginseng samples in tonghua city,jilin Province and huanren manchu autonomous county,liaoning province.3.Rapid and non-destructive cluster analysis were performed on the main root,powder and slices samples of American ginseng and dried ginseng with similar appearance but different efficacy.The model had an identification rate of 97% for American ginseng and 96% for dried ginseng.The main root of American ginseng samples had scores of-0.11581~-0.09378 on the first principal component(PC1)and scores of-0.20181~0.24027 on the second principal component(PC2),the scores on the third principal component(PC3)were-0.2136 ~0.27762.The main root of dried ginseng samples scored-0.12197~-0.093391 on the first principal component(PC1)and-0.13153~0.202 on the second principal component(PC2),the scores on the thirdly principal component(PC3)were-0.18039~0.32499.The classification rate of the verification set was 92%.The recognition rate of slices of samples of American ginseng and dried ginseng by the Fourier transform near-infrared non-destructive rapid qualitative clustering model were 98% and 100%,respectively.The slices of American ginseng samples had scores of 0.08163~0.11898 on the first principal component(PC1)and scores of-0.20351~0.21351 on the second principal component(PC2),the scores on the third principal component(PC3)were-0.26008~0.24875;The slices of dried ginseng samples scored 0.091448~0.118 on the first principal component(PC1),-0.16386~0.27606 on the second principal component(PC2),and the scores on the third principal component(PC3)were-0.19118~0.20145.The classification rate of verification set was 97%.The near-infrared non-destructive rapid qualitative clustering model had a recognition rate of 100% for both the American ginseng powder samples and the dried ginseng powder samples.The American ginseng powder samples had scores of 0.093576~0.10978 on the first principal component(PC1)and scores of-0.069762~0.19701 on the second principal component(PC2),on the third principal component(PC3)the scores were-0.15122~0.12728;The dried ginseng powder samples scored 0.1003~0.1226 on the first principal component(PC1)and-0.09364~0.20415 on the second principal component(PC2),on the third principal component(PC3),the scores were-0.11373~0.14514.The classification rate of the validation set in the model was 98%.The above results indicated that the main roots,slices and powder of American ginseng and dried ginseng samples with similar appearance can be clustered and analyzed by Fourier transform near-infrared fastly and non-destructively.4.For the four years old planted ginseng in the three places of tonghua city of jilin province,huanren manchu autonomous county of liaoning province and yichun city of heilongjiang province,tracing the the production area under non-destructive conditions.The established Fourier transform near-infrared non-destructive rapid clustering qualitative model had the recognition rates of 100%,97%,and 92% for the planted ginseng in tonghua city of jilin province,huanren manchu autonomous county of liaoning province,and yichun city of heilongjiang province,respectively.The planted ginseng samples from jilin province scored-0.11308~-0.090162 on the first principal component(PC1)and-0.26533~0.14477 on the second principal component(PC2).The scores on the third principal component(PC3)were-0.26913~0.33785;The planted ginseng samples of liaoning province scored-0.11607~-0.087543 on the first principal component(PC1),and the scores on the second principal component(PC2)were-0.1587~0.28846,the scores on the third principal component(PC3)were-0.19494~0.17396;The planted ginseng samples of yichun city of heilongjiang province scored-0.11952~-0.09351 on the first principal component(PC1)and-0.2517~0.21288 on the second principal component(PC2),the scores on the third principal component(PC3)were-0.22668~0.25716.The classification rate of the established model for the validation set was 82%.The above data indicated that the established model also need to constantly adjust the modeling set,and continuously to increase the more representative spectra in order to basically achieve the purpose of traceability of the garden ginseng in the three places of tonghua city of jilin Province,huanren manchu autonomous county of liaoning province and yichun city of heilongjiang province under non-destructive conditions... |
PDF Full Text Request