The Study On The Classification Method Of Organic Matter By Laser-Induced Breakdown Spectroscopy

Organic matter mainly refers to carbon-containing compounds composed of carbon,hydrogen and other elements,which are the material basis of life.They are widely present in people's production and life,including common food and drugs,biological tissues,agricultural products,and organic explosives.The rapid and accurate detection of organic matter can be applied to the identification of fake and inferior products,the traceability of the origin of agricultural products,and the early warning of explosives in anti-terrorism.Laser-Induced Breakdown Spectroscopy(LIBS)is an element analysis method based on the interaction between laser and matter.It is gradually being widely used for the advantages of real-time online,micro-loss,multi-element analysis and low cost.In the field of substance analysis.The analysis of material composition by LIBS technology mainly depends on the element composition of the sample.However,for organic matter,since the types of elements contained are very similar,mainly C,H,O,N,etc.,it is difficult to directly determine the type of organic matter by spectral.Therefore,with the help of chemometric algorithms,valuable information can be extracted from complex spectral data sets that cannot be intuitively discovered,so as to achieve accurate classification of organic matter.This article takes the classification of organic matter as the research objective,and mainly includes the following research contents.1.The experimental method was explored by classifying the leaves of three different plants.The characteristic line and the principal components of full spectral were selected as the data input of the classification algorithm,respectively.By comparing the classification results,PCA was determined as the preprocessing method of spectral data.The first several principal components were taken as the data input of LDA and SVM,and then the data analysis method of modeling and classification was carried out.The average classification accuracies of the two classification algorithms on the test set reached to 96.7%and 98.9%,respectively.2.The traceability experiment of fresh Ginkgo biloba leaves.The samples came from eight different locations in Xi'an.For the same location,100 Ginkgo leaves were collected from several trees.After the spectra were preprocessed by PCA,the training set and the test set were divided in a ratio of 7/3.The average classification accuracies of LDA and SVM on the test set samples can reach to 97.5%and 96.3%,respectively.3.Research on the identification of trace explosives.2-Nitroanisole,urethane and glycine with similar elemental composition were selected as the simulated samples of organic explosives,and black powder was used as inorganic explosives.Using alcohol and water as solutes,the samples were made into solution or suspension.Drop the prepared liquid sample on the aluminum plate to make the sample volume on the plate is 200?g/cm~2.After modeling and analysis,the average accuracies of LDA and SVM on the test set can reach to 96.7%and96.0%,respectively.The above results show that the combination of LIBS with these chemometric algorithms is an effective method for classification of organic substances,which is expected to be applied in the fields of fast traceability of food,in-situ identification of biological tissues,and remote analysis of organic explosives.