Abstract: To address the challenges in the loosening and conditioning process, where hot-air temperature settings rely heavily on operator experience, outlet tobacco temperature exhibits large fluctuations, and the non-steady-state stage is prolonged, this study investigates the modeling and optimization of the relationship between hot-air temperature and outlet tobacco temperature. Based on 170 sets of experimental data, bivariate linear regression and random forest algorithms were employed, combined with Riemann sum to quantify temperature deviations during the non-steady-state phase, to provide a basis for the optimization of temperature control of export tobacco leaves. Results showed that hot-air temperature and outlet tobacco temperature exhibit a significant positive correlation. Increasing the hot-air setpoint accelerates outlet temperature response and shortens the non-steady-state duration. The random forest model outperformed the linear model in fitting performance, achieving high prediction accuracy with a mean error of less than 0.2 ℃. The proposed “prediction & optimization” strategy contributes to enhance temperature control accuracy and stability in the conditioning process, providing a theoretical and practical foundation for digital and intelligent control in cigarette primary processing.