CAJ | 학술논문

　　目的实现羊奶中常用抗生素残留的快速检测。方法以电子鼻作为快速无损的检测仪器,将含有不同浓度的青霉素 G、硫酸庆大霉素、硫酸链霉素的羊奶作为研究对象,利用线性判别分析(linear discriminant analysis, LDA)和多层感知器神经网络(multilayer perceptron neural network, MLPN)等方法进行数据处理。结果LDA的判别结果总体上能够将不同浓度的抗生素区分开来,效果良好；MLPN的训练集和预测集的正确率也都在94.9%以上,取得了很好的预测效果。结论 LDA 判定分组和MLPN预测模型可以用于羊奶中不同浓度青霉素G、硫酸庆大霉素、硫酸链霉素的快速检测。
　　목적실현양내중상용항생소잔류적쾌속검측。방법이전자비작위쾌속무손적검측의기,장함유불동농도적청매소 G、류산경대매소、류산련매소적양내작위연구대상,이용선성판별분석(linear discriminant analysis, LDA)화다층감지기신경망락(multilayer perceptron neural network, MLPN)등방법진행수거처리。결과LDA적판별결과총체상능구장불동농도적항생소구분개래,효과량호；MLPN적훈련집화예측집적정학솔야도재94.9%이상,취득료흔호적예측효과。결론 LDA 판정분조화MLPN예측모형가이용우양내중불동농도청매소G、류산경대매소、류산련매소적쾌속검측。
Objective To develop a rapid detection method of usual antibiotics resedues in goat milk by electronic nose. Methods The goat milk with four concentrations (0, 50, 100, and 200μg/L) antibiotics, in-cluding penicillin, gentamycin and streptomycin was detected by electronic nose as a rapid non-destructive de-tecting equipment, to collect the volatiles emitted from the goat milk. Then the linear discriminant analysis (LDA) was applied to analyze the results and the response signals were chosen for the pattern recognitions by multilayer perceptron neural network (MLPN). Results LDA results showed that the response signals of the goat milk could distinguish the concentrations of antibiotics. The correct rates of the training group and the testing group of MLPNs were both more than 94.9%. Conclusion MLPN can be used as an effective fore-casting tool for predicting the concentration of three kinds of antibiotic in goat milk using the signals of elec-tronic nose.