Control of moisture content of rewarming equipment based on model predictive theory

Controlling the moisture content of drum warming and rewarming equipment is crucial for maintaining tobacco quality and ensuring stable downstream processing. The sieving and charging process presents a complex control challenge due to its multivariable, nonlinear nature with hysteresis and large inertia. Traditional PID + feedforward DCS systems fail to deliver optimal results. To address this, a predictive model based on XGBoost is developed for forecasting outlet moisture. A simplified Dynamic Matrix Control (DMC) algorithm using a single-weight approach is implemented on Siemens PLCs via Structured Control Language (SCL), tailored to the PLC's computational constraints. Field application in the Shijiazhuang Cigarette Factory demonstrates superior performance over existing methods without requiring extra hardware or communication modules. The controller dynamically adjusts its internal weighting in response to changing operating conditions, achieving robust adaptability and improved precision. Results confirm the system's effectiveness and stability.