Variation law and prediction model to determine the moisture content in tea during hot air drying

Abstract

AbstractTo study the changes in the moisture content of tea leaves during the hot air drying process of tea production, this study designed an experiment to monitor the dynamic change of moisture content of rolled green tea during the hot air drying process under different feeding amounts (800–1,200 g), drying temperature (90–120°C), and drum speed (20–30 rpm/min). We established a dynamic prediction model to determine the moisture content using the BP, Elman, and Particle swarm optimization Elman neural network (PSO Elman) algorithms. The prediction calculation was conducted by considering the drying temperature, drum speed, drying initial water, and prediction time as input and the water content prediction result during the tea drying process as the output. Additionally, we analyzed the significant factors to explore the dynamic changes in the water content of tea leaves under different drying conditions. Experimental results showed that the temperature, rotational speed, and feeding rate significantly affected the drying effect. Furthermore, we established a traditional multiple linear regression fitting model was established to compare and analyze with the above three neural network models. The verification and error analysis results of the four water prediction models showed that the PSO–Elman model could better predict the changes in the water content during the drying process.Practical ApplicationsThe change of moisture content in the process of tea hot air drying research can provide a theoretical basis for the hot air drying technology and process of tea, and have important significance for guiding tea processing and production, improving processing efficiency and tea quality.