The Research On Near Infrared Spectroscopy In The Extraction Process Of The Poria Cocos Polysaccharides

In recent years,traditional Chinese medicine(TCM)obtains more and more attention and shows superior development prospects with the continuous advancement of the modernization of TCM.However,The TCM preparations manufacturing process is complexity,and the dection technology mostly relies on the traditional analysis methods,lacking the effective process quality control technology.Therefore,there still need to try hard to promote the improvement in quality evaluation and control of TCM.The process analysis technology(PAT)is powerful and can realize real-time online monitoring of the TCM manufacturing process to ensure the quality of the product.Therefore,it’s necessary to promote the application of PAT on the TCM manufacturing process.The extraction process is one of the key part of the TCM manufacturing process,and the quality of the extract will directly affect the subsequent production.This paper took the application of near-infrared spectroscopy analysis technology and statistical process control methods in the extraction process of the polysaccharides of Poria cocos as research object.The main research contents and results were as follows.1.An analytical method based on near infrared spectroscopy(NIRs)was applied in the extraction process of the acid polysaccharides of Poria cocos.First,the extraction process lasted 6 hours and was repeated three times,and some certain extract liquid was taken out every 10 min.As a result,a total of 108 samples were collected.Then the NIR spectra was determined,and the content of the sample solution was determined by phenol-sulfuric method.Finally,PLSR was used to correlate the NIR spectra with the content to establish a quantitative model.In order to improve the models’prediction ability,various data division methods including Kennard-Stone(KS)、sample set partitioning based on joint x-y distances(SPXY)and the content descending,were discussed,and then the usable NIR spectral regions were selected.Several pretreatment methods,including Savitzky-Golay(SG)smoothing、multiplicative scatter correction(MSC)and standard normal variation(SNV),were also discussed in detail.Finally,Several variable selection methods,including genetic algorithm(GA)、particle swarm optimization(PSO)、competitive adaptive reweighted sampling(CARS)、monte carlo uninformative variable elimination(MCUVE)、random frog(RF)and bootstrapping soft shrinkage(BOSS),were carefully compared.As a result,the model established by means of the content descending for dividing the data,SG smoothing for pretreating the spectra and GA for selecting the wavelengths performed the best,where the coefficient of determination(R~2)was improved to 0.9618 and 0.9609 for calibration and prediction respectively,the RMSEP/RMSEC value was equal to 1.00 and the residual prediction deviation(RPD)was enhanced to 5.15.Such satisfying results clearly showed that the present model had a good prediction ability and thus was suitable for monitoring of the acid polysaccharides of Poria cocos during the extraction process.2.An analytical method based on NIR were investigated during the extraction process of the water-soluble polysaccharides of Poria cocos.The results of the second chapter showed the good applicability of NIR on the Poria cocos acid polysaccharides extraction process.Considering the content value of the water-soluble polysaccharides was much less than the acid polysaccharides,we went forword to investigate whether the NIR combined with multivariate calibration modeling methods were still applicable in the water-soluble polysaccharides.First,the extraction process lasted 2 hours and was repeated five times,and some certain extract liquid was taken out every 4 min and immediately collected 5 liquid samples in a row when reaching the end-point of the extraction.As a result,a total of 170 samples were collected.Then the NIR spectra was determined,and the content of the sample solution was determined by phenol-sulfuric method.And then the quantitative models were established between the NIR spectra and the content.The modeling methods and step was similar to the second chapter,several data division methods(KS、SPXY and the content descending),pretreatment methods(SG、MSC and SNV)and variable selection methods(GA、PSO、CARS and MCUVE)were used to improve the predictive ability of the quantitative model.Then the predictive ability of the quantitative model estabished by several multivariate calibration modeling methods,including PLSR、principle component regression(PCR)、support vector regression(SVR)、random forest(RF)and radial basis function(RBF)netural networks,was dicussed highlightly.As a result,the model established by means of the content descending for dividing the data,SNV smoothing for pretreating the spectra,GA for selecting the wavelengths and PLSR for the multivariate calibration modeling performed the best.The R~2 values were increased to0.8534 and 0.8480 for calibration and prediction respectively,the RMSEP/RMSEC value was equal to 1.00,and the RPD value was increased to 2.60.Although the PLSR model of the water-soluble polysaccharides was poorer than the the acid polysaccharides,which was the results that the lower content of water-soluble polysaccharides made the predictive aurracy of the model performed worse,the established model was still suitable for monitoring the extraction process of the Poria cocos water-soluble polysaccharides in summary.3.An analytical method based NIR combined with statistical process control(SPC)was applied in the extraction process of the Poria cocos water-soluble polysaccharides.A total of 25 end-point extract samples from 5 batches under the normal extraction manufacturing of the Poria cocos water-soluble polysaccharides collected from the third chapter were used to buid the control limit of the Shewhart、Hotelling T~2 and squared prediction error(SPE)control chart respectively.The real time release criteria of quantitative was established by the control limit of the Shewhart control chart with the control limit of the Hotelling T~2 and SPE control chart for the real time release criteria of quanlitative.And then the two criterias were tested by the 15 samples from 3 batches under the abnormal extraction manufacturing(abnormal solid-liquid ratio、extraction temperature and extraction time)and the 10 samples from 2 batches under the normal extraction manufacturing but not used for building the models.As a result,the real time release criteria of the Poria cocos water-soluble polysaccharides was the content values of 1.67-2.36%for the quantitative release criteria and the value of SPE less than 8.03 for the quanlitative release criteria.15 samples under the abnormal extraction manufacturing,and 10 samples under the normal extraction manufacturing all can be accurately identified according to the above two release criterias.Therefore,the results indicated NIR combined with SPC can be successfully applied to buid the real time release criteria of the extraction process of the Poria cocos water-soluble polysaccharides.