Prediction of chemical contents of cooking oil using near-infrared spectroscopy

Near-infrared (NIR) spectroscopy can be an alternative method that is faster and more accurate in monitoring the deterioration of cooking oil quality compared to chemical methods. This study aimed to assess NIR spectroscopy for monitoring the quality change of cooking oil that was used repeatedly, to determine the best spectra data pre-treatment and the number of main factors of partial least square (PLS) for prediction of the quality of cooking oil. Cooking oil samples were produced 0-15 times in the frying process using the double layer of the fryer so that there were variations in quality at each stage of frying with a total of thirty-two samples. Transflectance of cooking oil was measured by fourier transform NIR spectrometer at the wavelength of 1000-2500 nm. Peroxide number and free fatty acid content as the main chemical content of cooking oil parameters were determined by chemical method. Some spectra pre-treatments were applied and the calibration between NIR data and chemical data was carried out using PLS. The best calibration was selected based on the highest correlation coefficient (r), the lowest standard error of calibration (SEC), and the lowest standard error of prediction (SEP). The best accurate prediction of peroxide number was obtained by absorbance spectra with detrending pre-treatment at a PLS factor of 7 with a r value of 0.93, SEC of 0.17%, and SEP of 0.16%. Meanwhile, free fatty acid (FFA) content was original absorbance spectra at 8 factors of PLS with a r value of 0.94, SEC of 0.016%, and SEP of 0.019%. These results indicated that NIR spectroscopy can be used to monitor the quality change of cooking oil.