Abstract: In order to clarify the influence of different production processes on the stability of soybean oil, we used soybean oils obtained by pressing and extraction to deep-fat fry chicken fillets and we investigated the influence of frying time and temperature on the quality of the oil samples according to changes in acid value (AV) and carbonyl value (CV). Experimental results showed that AV changed only slightly with frying time. After being used at 220 and 240 ℃, the CV of pressed soybean oil was significantly higher than that of extracted soybean oil. Moreover, at 240 ℃, there were significant differences in both parameters (P < 0.05). Therefore, compared with the pressed oil, the extracted oil had better stability against frying at high temperature. At the same time, near-infrared (NIR) spectroscopy was used to predict the two stability indicators of soybean oil. A highly accurate predictive model was established using the characteristic bands selected by interval partial least square (IPLS) and successive projection algorithm (SPA). The AV prediction model had the best performance in the bands of 1 150–1 315 nm and 1 579–2 444 nm with a validation correlation coefficient of 0.955 and a root mean square error of prediction of 0.049. The CA prediction model had the best performance in the bands of 1 236–2 093 nm and 2 187–2 594 nm with a validation correlation coefficient of 0.946 and a root mean square error of prediction of 3.134. These results support that the accuracy of the model may be improved effectively by selecting the characteristic bands using IPLS and SPA.

Key words: pressed oil, extracted oil, acid value, carbonyl value, near-infrared spectroscopy, interval partial least squares, successive projections algorithm