| The quality of stable and controllable Chinese medicine is an important guarantee for the clinical efficacy of traditional Chinese medicine.During the storage process,traditional Chinese medicine is often affected by external conditions and its own characteristics,and often undergoes deterioration."Oil-releasing" is one of the common metamorphisms in the storage of traditional Chinese medicine decoction pieces.After the oil-releasing,the drug is lost and the effect is reduced.The severe one is toxic and life-threatening.Therefore,the rapid identification of the phenomenon of oil-releasing in the storage of traditional Chinese medicine,the objective judgment of the degree of oil-releasing and the real-time monitoring of the quality of the medicinal materials in the process of oil-releasing are important issues to be solved and studied.Armeniacae Semen Amarum is a commonly used bulk Chinese medicine,and it is also a typical representative of easy "oil-releasing" deterioration.In this paper,Armeniacae Semen Amarum was used as the research carrier.From the aspects of color,odor digitalization and information fusion,the discriminating model and quality prediction model of Armeniacae Semen Amarum were established.The main contents and results of the study are as follows:(1)Preparation of different oily samples and artificial sensory evaluation.In this paper,the samples of Armeniacae Semen Amarum with different oil-releasing levels were obtained by accelerated test and sample retention.The fuzzy mathematics method was used to comprehensively evaluate the appearance traits of Armeniacae Semen Amarum.According to the assessment results,Armeniacae Semen Amarum were divided into four different oil-releasing grades.(2)Determination of the chemical constituents in the Armeniacae Semen Amarum samples with different oil-releasing levels.As the degree of oil-releasing deepens,the content of amygdalin decreases,and the acid value and peroxide value increase gradually.The volatile components of Armeniacae Semen Amarum change after oil-releasing.Furfural and 2-bromopropiophenone may be the main components of Armeniacae Semen Amarum metamorphism.(3)Quality monitoring of Armeniacae Semen Amarum oil-releasing based on color digitization.In this paper,the color and profile color of Armeniacae Semen Amarum powder were determined by using Hitachi 3010 UV-Vis spectrophotometer and Konica Minolta CM-5 spectrophotometer.A discriminant model based on color digitization for different bitterness of Armeniacae Semen Amarum was established.Based on the colorization of powder color,the model established by Naive Bayes algorithm has a positive rate of more than 85%through the ten-fold cross-validation method and the external test set verification method.Based on the cross-sectional color digitization,the Logistic and Multiple Layer Perception algorithms have higher positive rates.Correlation analysis showed that there was a significant correlation between the intrinsic quality of Armeniacae Semen Amarum and the color of the powder and the color of the section.A predictive model of the powder color and intrinsic quality and a predictive model of the cross-section color and intrinsic quality were established.(4)Based on the odor digitalization of Armeniacae Semen Amarum oil-releasing quality monitoring research.In this paper,the Armeniacae Semen Amarum sample was subjected to odor determination using an a-FOX3000 odor fingerprint analyzer.Based on the odor digitalization,the high-temperature and high-humidity storage environment samples were selected to discriminate the degree of Armeniacae Semen Amarum oil-releasing by machine learning algorithm.The Logistic algorithm has the best recognition effect,which can better classify the different Armeniacae Semen Amarum.Correlation analysis showed that there was a significant correlation between the quality and the odor of Armeniacae Semen Amarum.A prediction model of amygdalin content based on odor information under defined storage conditions was established.(5)Monitoring of Armeniacae Semen Amarum oil-releasing quality based on color odor fusion information.In this paper,the data fusion method of feature variable fusion is used to combine the color information and odor information of Armeniacae Semen Amarum to obtain more comprehensive sensory information.A discrimination model of Armeniacae Semen Amarum oil-releasing degree based on color odor fusion information was established.Based on the color-odor fusion information of powder,Logistic,IBK,KStar,LMT and Random Forest algorithms have higher positive rates,which can better classify and distinguish Armeniacae Semen Amarum with different oil-releasing levels.Based on the cross-section color-odor fusion information,the positive rate of Logistic algorithm and KStar algorithm is higher,which can better classify and distinguish Armeniacae Semen Amarum with different oil-releasing levels.Correlation analysis showed that there was a correlation between the intrinsic quality of Armeniacae Semen Amarum and the fusion information.Establish a predictive model of fused information and intrinsic quality.(6)Comparison of different quality monitoring systems.In the discriminant model of Armeniacae Semen Amarum with different oil-releasing levels,the discriminative model is based on the color-odd fusion information of Logistic and IBK.The prediction model of amygdalin content in this paper,the best fit is based on the regression equation of section color eigenvalueor a regression equation based on the profile color-odor fusion information.The acid value prediction model constructed in this paper is based on the regression model of the powder color eigenvalues or a regression model based on powder color-odor fusion information.Studies have shown that it is feasible to monitor the quality of Armeniacae Semen Amarum oil-releasing based on the color,odor digitalization and information fusion.The established quality monitoring system can judge the oil-releasing levels of the Armeniacae Semen Amarum and predict the inherent quality. |
PDF Full Text Request