基于区间选择偏最小二乘法回归分析的煎炸老油鉴别模型

煎炸老油属于地沟油之一，长时间食用会对身体健康造成严重危害。本文首先对食用金龙鱼菜籽油过度煎炸10次，分别提取光谱信息中特征吸收峰（323、391、443nm）处的光谱吸收值。然后建立区间选择偏最小二乘法模型，提取最佳区间为5，最佳因子为7的波长区间，验证区间预测相关系数达到0．998。最后建立PKS回归方程Y1，Y2，Y3，通过交叉有效性验证Q22=-0．3461〈0．0975。提取成分t1、t2。分别得到回归方程的实际值与预测值相关系数为：323nm处，r=0．8478；391nm处，r=0．9754；443nm处，r=0．9285。

Identification model in frying oil based on interval partial least squares regression analysis

Frying oil belongs to one of the waste oils, and the long time eating does serious harm to health. First of all, the normal Arowana Rape Seed Oil is excessive frying for 10 times. And they were extracted spectral absorption values in the characteristic absorption peaks of 323 nm,391 nm and 443 nm from the spectral information. Then we establish method of interval estimation, and select partial least square model to extract the best region of 5, the best factor for 7 of the wavelongth interval. And then we verify whether the interval prediction correlation coefficient is 0. 998. Finally, the establishment of the PLS regression equation are Y1,Y2 and Y3. According to cross - validation, the verification Q22 is - 0.3461 less than 0.0975. Extracting components t1 and t2, we obtained correlation coefficients of the regression equation of actual and predicted value is 323 nm and r is 0.8478; 391nm and r is 0.9754; 443nm and r is 0.9285.