中文摘要：

新霉素在巯基乙醇存在的条件下与邻苯二甲醛生成的衍生物具有强荧光特性,可实现鸭蛋蛋清中新霉素残留含量的荧光测定。在模型建立过程中,分析了波长为280～390 nm光谱范围内的三维同步荧光光谱,确定检测鸭蛋蛋清中的新霉素含量的最佳波长差Δλ为110 nm;然后利用db10小波的2层分解对一阶导数同步荧光光谱进行去噪处理,并利用分段遗传算法（SGA）优选出了14个特征波长;最后应用最小二乘支持向量回归（LSSVR）建立了鸭蛋蛋清中的新霉素含量的预测模型,其模型预测集的决定系数（R2）和预测均方根误差（RMSEP）分别为0.9671和1.713。结果表明,SGA能有效提取出鸭蛋蛋清中新霉素对应的特征波长,有利于提高LSSVR模型预测精度,可实现鸭蛋蛋清中新霉素残留含量的快速测定。

英文摘要：

Derivate of neomycin and o-phthalaldehyde（OPA） has strong fluorescent characteristic under the condition of mercaptoethanol,and it can realize fluorimetric determination of neomycin content in duck egg white.In the process,synchronous fluorescence spectra from 280 nm to 390 nm were analyzed using three-dimensional synchronous fluorescence spectrometry,and the wavelength difference of 110 nm was selected to detect neomycin content in duck egg white.Then,the db10 wavelet with two levels decomposition was used to reduce the noise of first order derivative synchronous spectrum,and 14 characteristic wavelengths were obtained using subsection genetic algorithm（SGA）.Lastly,least squares support vector regression（LSSVR） was applied to build the prediction model of determination of neomycin content in duck egg white,and the determination coefficient（R2） and the root mean squared error of prediction（RMSEP） were 0.9671 and 1.713 respectively.This work proved that SGA could extract effectively characteristic wavelengths of neomycin in duck egg white,and it was helpful to improve the prediction accuracy of LSSVR model.The method could achieve the rapid prediction of neomycin content in duck egg white.