Authentication And Quality Evaluation Of Eucommia Ulmoides Leaves Based On Multi-information Fusion

The improvement of Chinese residents’ living standards has led to their increasing awareness of health care.Chinese medicine originates from life and plays an important role in the daily life of residents.Therefore,the traditional Chinese medicine equally used for food and medicine has attracted much attention.However,the quality of Chinese medicinal materials is affected by various factors such as the original plant,harvesting time,geographical origin,botanical part,etc.,resulting in uneven and unstable quality.Spectroscopy and chromatographic analysis techniques have been widely used in quality control of traditional Chinese medicine.Spectroscopy,as an environmentally-friendly green analysis method,combined with the content data of the characteristic components of chromatography,and supplemented by chemometric statistical methods,can provide a fast,effective and convenient technical reference for the quality control of traditional Chinese medicine.In this thesis,Eucommia ulmoides leaves are used as research objects.Ultravioletvisible spectroscopy(UV-Vis),Attenuate total reflection-Fourier transform infrared spectroscopy(ATR-FTIR),Near infrared spectroscopy(NIR),High performance liquid chromatography equipped with diode array detector(HPLC-DAD)were combined for analysis of E.ulmoides leaves from different origins.Combining with two unsupervised pattern recognition methods(Principal component analysis,Hierarchical cluster analysis),three supervised pattern recognition methods(Partial least squares discriminant analysis,Support vector machine,Random forest)were also used for the discrimination analysis.Support vector machine regression(SVMR)analysis was used for an in-depth analysis of different spectral and chemical characteristics of E.ulmoides leaves from the producing area.Finally,a systematic quality evaluation method for E.ulmoides leaves was established to provide a theoretical basis for the resources development.The results indicated that Genetic algorithm-support vector machine(GA-SVM)and Particle swarm optimization-support vector machine(PSO-SVM)combined with the mid-level data fusion strategy can effectively identify E.ulmoides leaves from different regions with 100% accuracy rate.Meanwhile,it was found that the effect of individual differences(male and female)of E.ulmoides on the accumulation of chemical constituents was less than that resulted by the planting area.RF models based on high-level data fusion from NIR and ATR-FTIR spectra are most suitable for identifying E.ulmoides leaves from different origins.The accuracy of the training set and the validation set were 92.86% and 93.44%,respectively.From the clustering tree diagram of HCA,it can be found that E.ulmoides leaves from Xinjiang Uygur Autonomous Region belonging to a single category regardless of the distance of the samples in the tree diagram,and it is the only one in all samples located in high latitudes and middle temperate zones.Seven components in E.ulmoides leaves were determined by HPLC-DAD method.The SVMR model was established combined with the spectral data for the effective content prediction of 7 active ingredients.The results showed that the traditionally central production areas(main production areas in Hunan,Guizhou,and Henan Provinces)can be used as suitable cultivation sites for E.ulmoides from perspective of these 7 active ingredients.Appropriate preprocessing methods can improve the results of predictive models.Max Ent model was used to simulate the potential distribution area of E.ulmoides The Wuling Mountains and Qinba Mountains are selected as highly suitable areas for E.ulmoides,which can be widely planted in the middle temperate zone to the south subtropical zone of China.Temperature and precipitation are the most critical factors affecting its geographical distribution.