Study on Detection and Classification of Tetracycline Residue in Duck Meat Using Synchronous Fluorescence Spectra and Support Vector Machine

Abstract

To the rapid detection of whether the tetracycline residues are excess in duck meat, the optimum characteristic wavelength difference λ was determined by synchronous fluorescence analytical method. The recognition model of different residual levels of tetracycline was established by using support vector machine classification algorithm. Firstly, the optimum wavelength difference λ for duck meat samples was determined as 70nm, and synchronous fluorescence spectra of different samples under the condition of λ 70nm were collected. Secondly, original synchronous fluorescence spectra were preprocessed by using standard normal variables change (SNV). Finally, 18 wavelength variables were selected from 121 wavelength variables of pretreatment spectra by using competitive adaptive reweighted sampling (CARS). Then the radial basis function (RBF) was selected as the kernel function of support vector classification (SVC), and the optimal kernel function factor C and g were determined as 2.83 and 1, respectively, which were obtained by using grid searching combined with 5-fold cross validation. The classification model of SNV-CARS-SVC was established, and the classification accuracy rate of the model was 95.7% for prediction sets samples. The results showed that the synchronous fluorescence analysis method could identify tetracycline different residual levels quickly and accurately, and a feasible method was provided for identifying the quality of duck meat.

Keywords

• : synchronous fluorescence spectra
• competitive adaptive reweighted sampling
• support vector machine
• tetracycline
• duck meat.