Establishment Of A Database Of Herbage Varieties Based On Terahertz Technology And Discussion On The Mechanism Of Their Patterns And Colors

Alfalfa pasture is a very important category of pasture species,and it plays a vital role in the development of China’s grass industry,animal husbandry,and dairy industry.However,due to the wide variety of alfalfa forages,and the appearance and size of seeds are extremely similar,it is extremely difficult to identify alfalfa seed varieties.The frequency of terahertz(THz)waves is between microwaves and mid-infrared waves,and has the characteristics of low energy,fingerprint and penetrability,so terahertz technology is widely used in the field of non-destructive testing.Through preliminary investigation and research,it is feasible to apply terahertz time-domain spectroscopy to the identification of alfalfa seed varieties.On this basis,this paper combines machine learning with terahertz time-domain spectroscopy technology to study the establishment of alfalfa seed terahertz spectrum database and the mechanism of alfalfa color.The research content and results of this topic are as follows:(1)Machine learning algorithm combined with terahertz time-domain spectroscopy technology to identify alfalfa seed varieties with small differences.Specifically,using terahertz time-domain spectroscopy(THz-TDS),four series of 11 alfalfa seed varieties were experimentally tested,and terahertz time-domain spectra and refractive index spectra in the effective frequency range of 0.2 to 1.0 THz were obtained.Through the qualitative analysis of the spectrum,it is found that the THz-TDS technology can distinguish different series of alfalfa seed varieties.Among them,the terahertz wave optical characteristics of different series of alfalfa seed varieties are quite different,while the same series of alfalfa seed varieties are different.The optical characteristics of the terahertz wave are relatively small.In order to improve the classification accuracy of alfalfa seeds,this paper uses genetic algorithm to optimize the BP neural network model(GA-BP)to model and calculate 440 sets of terahertz refractive index spectrum data of a series of alfalfa seeds with small differences.The calculation results are compared with the BP neural network model results.The results show that the BP neural network model optimized by genetic algorithm(GA-BP)has an average classification accuracy of 87.3% for alfalfa seed varieties,while the BP neural network model is 80.6%.Therefore,the GA-BP model can be used in the identification of alfalfa seed varieties with small interspecies differences,and at the same time provide a basis for the research on commercial identification of forage varieties.(2)Design and build a terahertz spectrum database system for alfalfa seed varieties.The designed alfalfa forage terahertz database system is mainly composed of four data tables,and the four data tables are respectively used to store administrator information,terahertz spectrum data information,alfalfa forage information and test information.Today,the terahertz spectrum data of 37 alfalfa varieties are stored in the database.The establishment of terahertz database of alfalfa seeds helps to provide a basis for systematic research on alfalfa seeds.(3)Using terahertz time-domain spectroscopy(THz-TDS),a total of 16 alfalfa seed varieties with three different colors: alfalfa,miscellaneous flower and yellow flower were tested,and the time-domain spectrum and absorption in the effective frequency band of 0.2-1.0 THz were obtained.Coefficient spectrum and other optical characteristic parameters,and on this basis,the qualitative identification analysis of 16 alfalfa seed varieties and their differences,and the mechanism of different colors were studied.Using cluster analysis(CA)to quantitatively calculate the Euclidean distance of the terahertz absorption coefficient spectrum obtained in the experiment,the results show that the alfalfa varieties with the same flower and color have a greater degree of similarity in the Euclidean distance,while the alfalfa species with different colors The calculated Euclidean distance is far from each other.Furthermore,this subject uses the BP neural network to calculate and analyze the content of nutritional components of three different color alfalfa seed varieties.The results show that different color alfalfa seed varieties have different nutrient composition ratio relationships.Among them,alfalfa’s protein: fat: water: ash: carbohydrate weight relationship is 0.173: 0.166: 0.221:0.224: 0.216,and alfalfa’s protein: fat: water: ash: carbohydrate weight relationship is0.174: 0.216: 0.198 : 0.233: 0.188,the weight relationship of the protein: fat: water: ash:carbohydrate of alfalfa is 0.199: 0.133: 0.276: 0.236: 0.165.This shows that the main factors for the different colors of alfalfa species are related to the nutrient composition content.In summary,this project uses terahertz time-domain spectroscopy technology and machine learning algorithms to effectively identify alfalfa seeds with small differences,initially establishes a terahertz spectrum database of alfalfa seeds,and discusses alfalfa forage.The reasons for the different colors lay the foundation for the in-depth study of the application of terahertz spectroscopy to the identification of forage varieties.