Research And Application On Rapid Identification Of Dendrobium Plants Based On Near Infrared Spectroscopy

Dendrobium officinale(Tie-pi-shi-hu in Chinese), ranked as the first of “Nine kinds of Chinese medicinal herbs”, is a precious traditional Chinese herbal medicine which belongs to Dendrobium species, and is valuable because of its excellent immunomodulatory effects and relative scarcity.Unfortunately, for this reason, a large number of Dendrobium devoninum Paxt.(Zi-pi-shi-hu in Chinese)and Dendrobium nobile Lindl.(Jin-chai-shi-hu in Chinese) imitated as D. officinale have entered the market. In addition, D. officinale from different origins exhibit considerable difference in its pharmaceutical efficacy and price. Hence, it needs to establish a rapid and absolutely non-destructive discrimination method to identify Dendrobium. In this study, micro-NIR spectrometer was used to identify three kinds of Dendrobium which are easily confused with each other and also Dendrobium officinale from seven different origins. Then, on this basis of identification, the internet of things system of Dendrobium identification was established.In this study, Near infrared discriminant model of three kinds of Dendrobium that easy to be confused were established both intact and powdered forms. The effect of sample size on spectroscopy and model was explored. The spectral reproducibility of samples passing 80 mesh and the prediction result of the model built by samples passing 80 mesh is better than that of 40 mesh and 60 mesh. The result indicated that the recognition rates and the rejection rates of discrimination models in identifying all samples were 100% both intact and powdered forms. Models with high predication stability and high precision can achieve rapid non-destructive identification of three kinds of Dendrobium.The identification between different origins of Dendrobium is more difficult than that between different species because the difference between origins is smaller than between species. Thus, at first,Near-Infrared Spectroscopy(NIR) combined with soft independent modeling of class analogy(SIMCA)was used to identify seven different origins of D. Officinale(sample form: powdered forms). The result indicated that recognition rates and rejection rates of discrimination models in identifying all samples were both 100% after the optimization of pre-processing and band selection and the main components.Models can identify D. Officinale from seven origins successfully. On this basis, three modeling methods that concluding SIMCA, PLS-DA and PCA-DA were used to identify D. Officinale of sevendifferent origins(samples form: intact samples) and the result is that the recognition rates is higher but the rejection rates is above 70% of SIMCA model, the correct rates is 70% of PLS-DA, while the result of PCA-DA is better with 94% correct rates.The construction of the internet of things system based on cloud computing can achieve a lot of functions such as spectrum collection of Near-Infrared Reflectance 、 spectrum comparison 、 model construction and predictive analysis. This system is more stability and accuracy with reasonable design and simple interface operation after the prediction and preliminary application of this system. The model development of Near-Infrared Spectroscopy of Dendrobium provided an effective and fast and non-destructive techniques for Dendrobium identification and research areas. The problem that Dendrobium can’t be quickly identified are resolved in China market.