Prediction of Shelf life of Vannamei Shrimp (Litopenaeus vannamei) Fillet in Freezing Conditions Based on Arrhenius Model and Artificial Neural Network

In order to the shelf-life prediction of white shrimp fillet (Litopenaeus vannamei) using Arrhenius mathematical model and Artificial Neural Network at different temperatures (-15, -25, -35, -45 °C), the qualitative changes of the fillet including salt extractable protein (SEP), K index, total volatile nitrogen bases (TVB-N), peroxide index (PV), barbituric acid (TBARS), electrical conductivity (EC) and sensory evaluation (SA) were investigated. For the Arrhenius model, the relative error range between the measured and predicted values for the quality factors i.e. TVB-N, SA, EC, TBRAS, K and SEP was -73.17-15.12, -2.54-13.04, -6.47-1.62, -0.81-0.00, -25.99-2.02, -5.59-0.82%, respectively. Regarding to Artificial Neural Network model, the relative error range between the predicted and measured values for the quality factors TVB-N, SA, EC, TBRAS, K and SEP were 0.00, 0.00, -0.38-0.00, 0.00, 0.00 and -0.08-0.03%, respectively. The MSE values of the Artificial Neural Network model were lower than the Arrhenius model in most of the qualitative factors. The R2 of the frozen shrimp quality factors of the Artificial Neural Network model was higher than the Arrhenius model, except for the SA factor. The artificial neural network model was able to better show the trend of changes in the quality of shrimp stored during the 6 months of the freezing period, at of -15 to -45 °C, compared to the Arrhenius model.