Modeling and Predicting the Oxidative Stability of Olive Oil during the Storage Time at Ambient Conditions Using Artificial Neural Network

Oxidative stability is one of the most significant parameters in maintaining the quality of olive oil during the storage time. The confidence of the stability and quality of olive oil is a great concern for producers and sellers. Therefore, this study aimed at modeling of the oxidation stability of olive oil by using artificial neural network (ANN) in order to improve the quality control process of this product. In the present study, a Feed-forward Neural Network)FF-ANN (was used to estimate the oxidative stability of olive oils during storage. In the neural network structure, the parameters of acidity, peroxide value (PV) specific extinction coefficient K232, phenolic compounds, structure of saturated and unsaturated fatty acids were used as input variables, and the extinction coefficient k270 was used as the output variable. The Feed-Forward-Back-Propagation network using the Tangent Sigmoid transfer function, Levenberg–Marquardt learning algorithm, and ten neurons in the hidden layer with lowest mean square error, and the best regression coefficient was determined as the best neural model. The regression coefficients of the best FF-ANN model in (30-120-210-300-420) days were 0. 936، 0. 955, 0. 957, 0. 974 and 0. 9769, respectively and the mean square errors were 0. 0057, 0. 0015, 0. 0012, 0. 0046, and 0. 0062, respectively. Our analysis demonstrated that FF-ANN was a powerful tool capable to predict oxidative stability of olive oils during the storage time.