Modeling of Oxidation Stability of Canola Oil Using Artificial Neural Networks during Deep Fat Frying of Potatoes

Abstract

The objective of this study was to apply artificial neural network (ANN) modeling to predict the physicochemical properties of canola oil during deep fat frying of potatoes. The technique was used for testing a 40-h period of deep fat frying at temperatures of 150, 165 and 180C with and without the presence of tert-butylhydroquinone (50 and 100 ppm). Monitoring was done on the speed at which the oil samples were destroyed by taking evaluations for oxidation parameters: peroxide value (PV), carbonyl value (CV) and total polar compounds (TPCs). Inputs to the network were the parameters of time, temperature and antioxidant concentration, and oxidation parameters were the output. Results show that data predicted by the models were in good agreement with experimental data (r = 0.959, r = 0.969 and r = 0.988 for PV, CV and TPC, respectively). Furthermore, parameters of time and temperature were determined as the most sensitive inputs of the best ANNs for predictions of the PV and CV, respectively. Furthermore, parameters of time and temperature were determined as the most sensitive inputs of the chosen ANNs for the prediction of the TPC. Practical Applications This article aims to investigate the effect of time and temperature on canola oil oxidation parameters during deep fat frying of potatoes; the effect of the presence and absence of antioxidant (tertbutylhydroquinone) on the quality of canola oil during deep fat frying of potatoes; and estimation of canola oil quality during deep fat frying of potatoes through artificial neural network models.