Storage quality prediction of winter jujube based on particle swarm optimization-backpropagation-artificial neural network (PSO-BP-ANN)

The quality prediction of winter jujube during storage is crucial to improve its postharvest value. Postharvest winter jujube was treated with 1 μL L−1 1-MCP and 5 μL L−1 O3 respectively. 22 physiological and biochemical indexes were determined at different storage periods, and comprehensive quality measurements were obtained after the data were combined and weighted. Correlation analysis showed that total flavonoids, b* value, c* value, total phenolics, healthy fruit rate had the highest correlation coefficient with the comprehensive quality evaluation value of winter jujube. Single BP and optimized BP (GA-BP and PSO-BP) neural networks of 3 treatments were used to predict the storage quality of winter jujube. The results showed that PSO-BP had the highest prediction accuracy and the overall fitting rate was above 95 %, indicating that the established PSO-BP model could effectively predict the storage quality of winter jujube after harvest.