PSO-based BP-ANN predictive model of S. Typhimurium in processing of surimi with citric acid

Foodborne pathogenic contamination is a major problem of surimi production. In this study, the effect of variables, namely citric acid concentration (0.5, 1, and 2%), process temperature (4 and 25 degrees C) and time (1-15 min) on the inactivation of Salmonella Typhimurium (S.Typhimurium) were investigated. The results indicated that citric acid had a significant effect on the survival of S.Typhimurium. To describe the kinetics of S.Typhimurium, both back-propagation artificial neural network (BP-ANN) and particle swarm optimization-based back-propagation artificial neural network (PSO BP-ANN) were used to develop models for simulating the dynamic population of S. Typhimurium. The novelty of this work consisted in the application of combining PSO and BP-ANN together as an optimization strategy to enhance its predictive ability. The results of the new model with PSO algorithm suggested a more accurate prediction model. With the optimal ANN-PSO model, the coefficient of determination (R-2) were 0.9786 and 0.9985; mean squared error (MSE) values were 0.0499 and 0.0049 for the training and testing data set, respectively. Practical applicationsSurimi is an important intermediate product in Asia, while fresh surimi may be contaminated with Salmonella Typhimurium during processing, transportation, and storage. In this study, Citric acid has been used to control microbial growth, and extend food shelf-life. PSO-based BP-ANN model were selected to predict the population of Salmonella Typhimurium in processing of surimi with citric acid as a more accurate prediction model. As surimi is constantly consumed, predicting and monitoring the harmful pathogenic bacteria is vital to ensure the food safety.