摘要
应用同步荧光光谱技术结合分段连续投影算法(SSPA)实现了对鸭蛋蛋清中强力霉素含量的快速检测。首先对蛋清、强力霉素标准品和含有强力霉素的蛋清进行三维同步荧光光谱分析,确定了其最佳波长差(△λ)为110 nm;然后对60个含有强力霉素的蛋清样本进行同步荧光光谱扫描,运用SSPA分别从全光谱和划分不同子区间中提取特征波长,作为多线性回归(MLR)、偏最小二乘法(PLS)、主成分回归(PCR)的输入变量,分别建立了SSPA-MLR、SSPA-PLS和SSPA-PCR模型。结果表明,利用SSPA法可以压缩模型的输入变量,缩减率为90.8%。划分为6个光谱子区间的SSPA-MLR模型的预测结果最好,其预测集样本的决定系数(R2)和预测均方根误差(RMSEP)分别为0.9896和0.89,有效地提高了模型的效率和预测能力。该方法为快速检测鸭蛋蛋清中抗生素提供新的方法。
In order to carry out the determination of doxycycline content in duck egg white rapidly, synchronous fluorescence spectroscopy combined with section successive project algorithm （SSPA） was applied. Firstly, the egg white, doxycycline standard solution and the sample of egg white containing doxycycline were analyzed by 3D synchronous fluorescence spectroscopic method, and 110 nm was determined as the optimum wavelength difference （ Δλ ） ; then spectra of sixty samples were obtained, characteristic wavelengths of full wavelength and different subintervals were chosen by SSPA, and characteristic wavelengths were used as input variables of MLR, PLS and PCR model, respectively. Finally, the SSPA-MLR, SSPA-PLS and SSPA-PCR model were established. The results showed that application of SSPA method could reduce the model input variables, and the reduction rate was 90. 8%. The forecasting results of SSPA-MLR model from six spectroscopic subintervals were best, and the determination coefficient （R2） and the root mean squared error （RMSEP） were O. 9896 and 0. 89, respectively, so it could improve the efficiency of the model and predictive power. A new method for rapid determination of doxycycline in duck egg white was provided.
出处
《分析试验室》
CAS
CSCD
北大核心
2013年第7期10-14,共5页
Chinese Journal of Analysis Laboratory
基金
国家自然科学基金项目(31101295)
新世纪优秀人才计划(NCET-09-0168)
国家高技术研究发展计划(863计划,2008AA10Z209)
江西省教育厅科技计划项目(GJJ12244)资助
