Genetic algorithm combined with back-propagation artificial neural network (GA-BP-ANN) to optimize ultrahigh pressure Actinidia arguta juice sterilization conditions to extend shelf life.

In this study, response surface methodology and GA-BP-ANN model were used to optimize the process conditions for ultrahigh pressure (UHP) killing of E. coli in Actinidia arguta juice. The results showed that the GA-BP artificial neural network model predicted the optimal UHP sterilization conditions (holding time 17 min, 350 MPa, and 2 times) with an E. coli death order of magnitude of 7.56. Both UHP and pasteurization had better sterilizing efficacy and mitigated the changes in pH and TSS values of the juice during storage. UHP sterilization preserved the polyphenols, flavonoids and ascorbic acid content of the juice better. A first-order kinetic model predicted that the UHP group could store Actinidia arguta juice at 4 °C for 95 days, whereas the untreated juice could only be stored for 21 days.