Abstract: Aiming at the problems of large parameter fluctuation, untimely control and high energy consumption of the drying machine, combined with the tobacco production process, the influencing factors of moisture content at the entrance of the drying machine were analyzed, and the variation laws of temperature and humidity in the external environment and the production workshop were studied. The experimental study of moisture content of cut tobacco was carried out under different temperatures and humidity conditions, and the influence laws of temperature and humidity changes in the process environment on the moisture content of cut tobacco were revealed. Based on the prediction model of multi-source environmental temperature and humidity data fusion and BP neural network, a dynamic prediction mathematical model of moisture content of the tobacco drying process was established. Through the training and optimization in production practice, the real-time prediction of inlet moisture content of the tobacco drying process was realized. The results showed that the change of ambient temperature and humidity was an important factor affecting the moisture content of cut tobacco, and the influence was the largest when the ambient temperature was higher than 20 ℃ and the ambient humidity was higher than 50%. The model could accurately predict the inlet moisture content of the tobacco drying process in real time, and the maximum error was controlled within ±1.5%. The maximum values of MAE and RMSE were 0.80% and 0.917 1%, respectively. The predicted values met the requirements of production practice, which had important guiding significance to ensure the stability of product quality and the accurate control of process.

Key words: cut tobacco drying, moisture content, environmental impact, temperature, humidity, prediction model, neural network