【摘要】：發(fā)展了一種基于近紅外自相關(guān)譜定性定量分析摻三聚氰胺奶粉的檢測方法。分別配置40個純奶粉樣品和40個不同質(zhì)量百分比濃度的摻三聚氰胺奶粉(10-4%~40%,w/w)樣品,采集了所有樣品的一維近紅外漫反射光譜,以奶粉中摻入的三聚氰胺濃度為外擾進行相關(guān)計算,選擇隨濃度變化敏感的7 000~4 200cm-1為建模區(qū)間。在提取自相關(guān)譜信息的基礎(chǔ)上,建立了定性定量分析摻三聚氰胺奶粉的偏最小二乘模型,并與常規(guī)一維近紅外譜模型的預(yù)測結(jié)果進行了比較。所建立的方法對未知樣品的識別正確率為100%,預(yù)測均方根誤差(RMSEP)為0.63%;而一維近紅外譜的識別正確率為96.2%,RMSEP為0.84%。研究結(jié)果表明:相對于常規(guī)一維近紅外譜,所建立的方法能提供更好的預(yù)測結(jié)果,其原因可能是自相關(guān)譜能提取更多的特征信息。
[Abstract]:A method for qualitative and quantitative analysis of melamine-doped milk powder based on near-infrared autocorrelation spectroscopy was developed. Forty samples of pure milk powder and 40 samples of melamine-doped milk powder (10-4%) with different mass percent concentrations were prepared respectively, and all samples were collected from 1-D near-infrared diffuse reflectance spectra. The concentration of melamine adulterated in milk powder was used as the external disturbance to calculate the correlation. The sensitive 7 000 200cm-1 was selected as the modeling interval. On the basis of extracting autocorrelation spectrum information, a partial least square model for qualitative and quantitative analysis of melamine-doped milk powder was established, and the prediction results were compared with the conventional one-dimensional near infrared spectrum model. The accuracy of the proposed method is 100 for unknown samples, and 0.63 for root-mean-square error (RMSEP), while that for one-dimensional near-infrared spectrum is 96.2and 0.84. The results show that the proposed method can provide better prediction results than the conventional one dimensional near infrared spectra, which may be due to the fact that autocorrelation spectra can extract more feature information.
【作者單位】： 天津農(nóng)學(xué)院農(nóng)學(xué)與資源環(huán)境學(xué)院;天津農(nóng)學(xué)院農(nóng)業(yè)分析測試中心;天津農(nóng)學(xué)院工程技術(shù)學(xué)院;
【基金】：天津市科技特派員項目(16JCTPJC47500) 國家自然科學(xué)基金基金項目(31201359,81471698) 天津市自然科學(xué)基金項目(14JCYBJC30400,14JCYBJC43700)資助
【分類號】：O657.33;TS252.51
，

本文編號：2307537