Rapid Discrimination And Analysis Recycled Oil By Chemometrics Combined With Infrared Spectroscopy

40 qualified edible oils and 44 recycled oils were collected and analyzed. 25 qualified edible oils and 39 recycled oils were selected to compose training set. The best classification result is coming out with three misclassified with principal Component analysis. Principal component analysis(PCA) was used to compress thousands of spectral data into several variables and describe the body of spectra, the analysis suggests that the accumulate reliabilities of PC1, PC2 and PC3(the first three principle components) are almost 95% and corresponding1743~1710cm-1&1172~1130cm-1,2945~2844cm-1&1728~1689cm-1,2987~2840cm-1& 1731~1660cm-1 are the most richest band for spectral information of samples. The training set was used to build 2 discrimination analysis(DA) models. the one uses the original spectra combined with full spectral band,the other one uses original spectra combined with most richest information band. the classification result of all are one sample misclassified. but the model uses original spectra combine with most richest information band need to process less data, that means the processing speed is increased. Using two batches of samples provide by Ministry of Health to established DA model and combine with original spectra and most richest information band. Using this model to classify the 247 samples from Guang Dong province, the result shows the correct recognition rate of 73.7%, which shows this method can be used to distinguish qualified edible oil rapidly and relaible,accurately and soundly.Using Fourier transform infrared spectroscopy measured the infrared spectra of the samples, Using high temperature gas chromatography measured diglyceride,triglyceride and the content of free fatty acid index proportion of samples,try to establish quantitative analysis model of partial least squares with different spectral preprocessing and number of factors, the cross check results show that the FFA、MG、DG and TG index coefficient R2 value was 0.86723,0.69144,0.87177,0.86266, respectively, the indexes of cross check all RMSEC of variance valueswere 0.162, 0.0569, 0.345, 0.403.