Identification Model in Frying Oil based on Interval Partial Least Squares Regression Analysis

摘要

As we known frying oil belongs to waste oils when it has been excessive used, long time usage also cause serious effect. This paper chooses dragon fish oil which was fried 10 times excessively. We can extract the characteristic in absorption peak (323.391.443nm) of spectral absorption value as the dependent variable. Then build the interval partial least square model, Through the MATLAB, we can extract the optimum interval is 5 and the best factor of wavelength range is 7. Prediction of correlation coefficient for R is 0.998. By the cross validation verification Q_(22)=-0.3461<0.0975, we can get the establishment of PLS equation as Y_1, Y_2, Y_3. The model which we build can predict the content situation of characteristic absorption peak in frying oil effectively.