GA‐ANN and ANFIS Models and Salmonella Enteritidis Inactivation by Ultrasound

Abstract

AbstractIn this study, adaptive neuro‐fuzzy inference system (ANFIS) and genetic algorithm–artificial neural network (GA‐ANN) models were used to predict inactivation of Salmonella enteritidis by ultrasound. The effect of amplitude levels, duty cycles and irradiation time of ultrasound on inactivation of S. enteritidis was investigated. The inactivation rate of S. enteritidis was increased by intensifying the amplitude levels and the best inactivation was achieved at 37.5 μm amplitude that S. enteritidis population was reduced to 1.67 cfu/mL. The high inactivation rate of S. enteritidis was achieved under duty cycle of 0.7:0.3, with reduction of population to 1.02 cfu/mL. The overall agreement between ANFIS predictions and experimental data was also very good (R2 = 0.974). The developed GA‐ANN, which included 12 hidden neurons, could predict S. enteritidis population with low mean squared error, normalized mean squared error and mean absolute error equal to 0.083, 0.023 and 0.200, respectively. The results indicated that both GA‐ANN and ANFIS models could give good prediction for the population of S. enteritidis. Sensitivity analysis results showed that ultrasound time was the most sensitive factor for the prediction of S. enteritidis population.Practical ApplicationsMicrobial inactivation by thermal treatment does damage to the organoleptic properties of foods. However, using ultrasound (US) for microbial inactivation minimizes this change. Generally, the advantages of US over thermal treatment are minimal flavor loss, significant energy saving and greater homogeneity.