Research On The Terahertz Spectral Database Of Common Antibiotics In Agriculture

China is a major country in the production and use of antibiotics.Every year a large amount of antibiotics are added to the feed used as growth-promoting agents in animal husbandry,and the problem of antibiotic residues is ultimately harmful to human health.In this work,12 common antibiotics used in agriculture were taken as the targets of interests,combining with terahertz spectroscopy and database technology,the research on the terahertz spectral database of antibiotics was carried out.In the research,the norms of sample preparation and spectra collection were drawn up,and terahertz spectral database system was designed and developed.Relying on the spectral database system,three similarity algorithms were tested in spectral matching experiment,and a new spectral matching mode named characteristic peak region matching was presented after the analysis on the matching process,whose optimization in spectral matching got verified through experiments.Main research contents and results are summarized as follows:(1)Basing on the previous research and relevant experimental experience,the norms of sample preparation and spectra collection got drawn up.The norm of sample preparation includes material choosing,pretreatment,parameters of sample preparation and so on.The norm of spectra collection includes experimental environment,spectrometer configuration,personnel practice and soon.(2)The terahertz spectral database system was designed and developed.Spectral database system was divided into database and database operating software.The database was constructed with three tables,that was used for separately storing the data of administrator information,detecting parameter and spectra;The operating software with many functions,like spectral matching and data updating,support the data read-write and save in the form of csv,xlsx and txt,and users can access the database via local area network.(3)three similarity algorithms(spectral angle,Pearson and Jaccard)were studied in spectral discrimination.Three algorithms were used to discriminate the type of antibiotics by calculating the similarity of spectra,and the discrimination rates of three algorithms were 100%,91.7%and 100%respectively,which revealed the great potential of the algorithms in spectral discrimination area.Compared to Pearson algorithm,the values of similarity of the spectra from same antibiotic,were smaller by the two algorithms of spectral angle and Jaccard.The three algorithms can't discriminate the type of samples clearly,and the matching progress need to be discussed and optimized.(4)A new spectral mode named characteristic peak region matching was adopted to optimize the spectral matching.In the view of the deficiencies in continuous region matching,characteristic peak region matching was presented.According to the matching results,all the discrimination rates of three algorithms were 100%.Compared to continuous region matching mode,characteristic peak region matching mode increased the values of similarity of the spectra from same antibiotic,enhanced the discrimination capacity of the algorithms obviously and realized the optimization for spectral matching.