| Poria cocos,as a medicinal and edible medicinal material,has a variety of biological properties such as immune regulation,anti-oxidation,and anti-tumor.It has high medicinal and edible value.It is widely used in Chinese medicine prescriptions and Chinese patent medicines,as well as in health.Modern studies have shown that Poria cocos mainly contains polysaccharides,triterpenes,proteins and other chemical components,among which polysaccharides are the main active components.The 2020 edition of the Chinese Pharmacopoeia only stipulates items such as thin-layer identification and moisture inspection for Poria cocos,without any content determination items.Therefore,it is still particularly necessary to improve the evaluation and control methods of Poria cocos quality.This thesis is based on near-infrared spectroscopy combined with chemometrics technology to quickly determine the content of Poria cocos and its antioxidant activity.At the same time,the origin and adulteration of Poria cocos were identified and analyzed,and a robust and reliable qualitative and quantitative model was established to achieve the purpose of quickly evaluating the quality of Poria cocos from various aspects.The completed research content and results of this thesis are as follows:1.Rapid determination of the content of Poria cocos polysaccharides:This study collected 71 batches of Poria cocos from different origins and different batches,and used the phenol concentrated sulfuric acid method to accurately determine the polysaccharide content,and correlated it with the near-infrared spectroscopy data of the sample to establish Quantitative model.The study compared the standard normal transformation（SNV）,Smooth,multivariate scattering correction（MSC）,Derivative and their combination,and compared genetic algorithm（GA）,particle Group optimization algorithm（PSO）,competitive adaptive re-weighting sampling method（CARS）,finally established the optimized partial least squares（PLS）quantitative model,and determined the coefficient（R2）,Root Mean Square Error（RMSE）,Mean Absolute Error（MAE）and Residual Predictive Deviation（RPD）multi-index evaluation model performance.The results of the optimized 1D9S-GA-PLS quantitative model are:the R2values of the training set and the prediction set are both greater than 0.95,and the RPD value is 4.86.The results show that the quantitative model established in this paper can quickly and accurately predict the polysaccharide content of Poria cocos.2.Rapid determination of the antioxidant capacity of Poria cocos:DPPH,ABTS and FRAP methods were used for all batches of Poria cocos extracts to determine the DPPH free radical scavenging ability,ABTS free radical scavenging ability,and Fe3+reduction ability.Correlate the measured data with near-infrared spectroscopy to establish three models based on near-infrared spectroscopy to quickly determine the antioxidant capacity of Poria cocos.In order to optimize the model,pretreatment methods such as SNV,MSC,1D5S,1D9S,2D5S,2D9S,and variable selection methods such as GA,CARS,and PSO were investigated.The results showed that the R2values of the training set and prediction set of the three GA-PLS quantitative models were both greater than 0.94,and the RPD values were both greater than 4.00.The results show that the PLS optimization models established in this paper have good predictive performance and can quickly determine the antioxidant capacity of Poria cocos.3.Identification and detection of Poria cocos powder adulteration:Mix10 batches of Poria cocos powder into corn starch and flour in different proportions（5%-50%）to prepare adulteration samples;then collect near-infrared spectroscopy information for qualitative and quantitative analysis.When establishing of the qualitative model,different modeling methods were compared,including partial least square discrimination（PLS-DA）,support vector machine（SVM）,linear discrimination（LDA）.And7 preprocessing methods and 7 feature selection method were compared.Then the discrimination error rate（ERR）,the true positive rate（TPR）,the true negative rate（TNR）and the F1 value were used as evaluation indicators.The results showed that the optimized DMSC-relieff-PLSDA qualitative model can accurately identify the three types of Poria cocos samples（ie,pure Poria cocos powder,adulterated corn starch,and adulterated flour）,and the ERR is all 0.The established DMSC-CARS-PLS quantitative model was used to quickly determine the ratio of adulterated flour and starch in Poria cocos.The R2values of the model results were all greater than 0.99,and the RPD values were all greater than 34.The results show that near-infrared spectroscopy combined with chemometrics can establish a reliable qualitative and quantitative model,and successfully realize the identification and detection of adulteration of Poria cocos powder. |
