| In recent years, the economy of China has made rapid development, people’sliving standard has been greatly improved, and the health problems caused by foodsafety have become the focus of attention of the whole society. Therefore, we need arapid and easy detection method to detect the presence of harmful substances in food.In this study, we use the fluorescence analysis method for the fast detection ofaminoglycoside antibiotic residues in Jiangxi duck.First of all, fluorescence spectroscopy combined with successive projectionalgorithm (SPA) was investigated for the fast determination of kanamycin residuesin duck. Research of the fluorescence spectrum of kanamycin standard samplesolution and the duck samples which contained kanamycin, discusses the reactionconditions and spectral characteristics, analysis of the fluorescence spectrum data ofkanamycin standard samples and duck samples, the best excitation wavelength ofkanamycin was330nm, and the emission wavelength was440nm. Successiveprojection algorithm was applied for the extraction of characteristic wavelength from257variables; the support vector machine (SVM) algorithm was developed for theprediction of kanamycin residues in duck. Then the MLR, PLS and GA-SVMmodels were developed for performance comparison. The determination coefficient(R~2) and root mean square error of predict (RMSEP) were used as the modelevaluation indices. The results indicated that the best performance was achieved bySPA-SVM model with R~2=0.9912and RMSEP=0.374ug/kg. An excellent predictionprecision was achieved. According to the detection of duck samples, the detectionlimit was150ug/kg. In conclusion, successive projection algorithm was a powerway for characteristic wavelength selection, and it was feasible to determine thekanamycin residues in duck, and an excellent prediction precision was obtained. Thisstudy supplied a new and alternative approach to the further appliation offluorescence spectroscopy in the fast determination of aminoglycoside antibiotics.Secondly, fluorescence spectrum combined with chemometrics methods wasapplied for turbidity determination of residual quantity of gentamicin in duck. Firstly,using the orthogonal experimental design method to design the conditions that thegentamicin standard solution could generate fluorescence, then the best conditionswas:2.0mL o-phthalaldehyde,3.0mL mercaptoethanol,0.5mL emulsifier OP-10and20min reaction time; secondly, data of duck samples were collected usingfluorescence spectrophtometer equipment, the detection limit was50ug/kg; lastly,models with predictive residual quantity of gentamicin in duck using PLS was established in the whole wavelength range. The results showed that, after smoothing,the models by center method was the best, the RMSECV and RMSEP were0.254ug/kg and0.0711ug/kg respectively, and the calibration and prediction correlationcoefficient(r) were0.9585and0.9976. Using fluorescence spectrum for rapiddetection of residual quantity of gentamicin in duck was feasible.Finally, use the response surface method to optimize the experimentalconditions, and the results ware: the fluorescence intensity of neomycin sulfatestandard solution was associated with o-phthalaldehyde, mercaptoethanol, emulsifierOP-10and reaction time. The optimum reaction conditions of this experiment were:1.5mL o-phthalaldehyde,1.0mL mercaptoethanol,2.0mL emulsifier OP-10and77min reaction time. Acquisition of the spectral data of60duck samples in the rangeof340~600nm. Genetic algorithm was used to model the amount of fluorescencespectra of neomycin residue in duck samples, and the prediction set was predicted bythe sample. The prediction correlation coefficient was0.9949. Research showed thatthe method was reasonable. The study supplied a new and alternative approach to thefurther appliation of fluorescence spectroscopy in the fast determination ofaminoglycoside antibiotics. |
PDF Full Text Request