BACKROUND: Apple p omace, o range peel and p otato p eel, which h ave important antioxidative compounds in their structures, are b yproducts obtained from fruit o r vegetable processing. Use of vegetable extracts is p opular and a common technique in the p reservation o f vegetable oils. Utilization o f apple p omace, o range peel and p otato p eel extracts as n atural antioxidant agents in refined sunflower oil during storage in order t o r educe o r r etard o xidation was investigated. All b yproduct extracts were added at 3000 ppm to sunflower oil and diff erent nonlinear models were constructed for t he estimation of oxidation parameters.

RESULTS: Peroxide values of sunflower oil samples containing d ifferent natural extracts were f ound to be lower compared to control sample. Adaptive neuro-fuzzy inference system ( ANFIS) and artificial neural networks (ANN) were used for t he construction o f models that could p redict the o xidation p arameters and were compared to multiple linear regression (MLR) for t he determination o f t he best model with h igh accuracy. It was shown t hat t he ANFIS model with h igh coeffi cient o f determination (R2= 0. 999) performed b etter compared to ANN (R2= 0. 899) and M LR (R2= 0. 636) for t he prediction ofoxidation p arameters

CONCLUSION: Incorporation of diff erent n atural byproduct extracts into sunflower oil p rovided an important retardation in oxidation during storage. Eff ective predictive models were constructed for t he estimation of oxidation p arameters u sing ANFIS and A NN modeling t echniques. These models can b e u sed t o p redict oxidative parameter values.