Abstract
To predict the oxidation parameters of soybean oil (SBO), we utilized five levels of black plum peel extraction (BPPE) antioxidant concentration (0, 400, 800, 1200, and 2000 ppm) and four levels of oil storage time (0, 8, 16, and 24 days) under accelerated oxidation conditions (temperature 60 °C). We investigated the peroxide value (PV), thiobarbituric acid (TBA) value, acidity, conjugated diene (CD) content, and carbonyl value (CV). Artificial neural networks were employed using MATLAB software for prediction. Several feed-forward back-propagation networks with 2-6-5 topologies were examined, achieving correlation coefficients greater than 0.959 and mean square errors (MSE) < 0.009. The optimal model utilized a sigmoid logarithm activation function, a jumping learning pattern, and 1000 learning cycles. These models demonstrated high correlation coefficients (above 0.912) in predicting the oxidation process of SBO.
doi: 10.17756/jfcn.2024-185
Citation: Mohammadi-Moghaddam T, Kariminejad M, Bakhshabadi H, Taghavi E, Morshedi A. 2024. Comparison of Sigmoid Logarithm and Hyperbolic Tangent Functions in Modeling the Oxidation Parameters of Soybean Oil Containing Extract of Black Plum Peels Natural Antioxidant. J Food Chem Nanotechnol 10(3): 134-139.
Downloads