Design of the model predictive control system for paddy circulation counter-flow drying

Drying is one of the most important processing tasks in modern agriculture. However, the existing circulating dryers have been limited to manual operation and empirical experience, leading to suboptimal drying efficiency and low product quality. Furthermore, the paddy deep-bed drying can also exhibit strong nonlinear dynamics, time delays in the heat and mass transfer, and multivariable coupling among key operational parameters, such as the grain moisture content, airflow rate, and drying temperature. Particularly, there is a high sensitivity to multiple internal and external disturbances, including the fluctuations in the ambient humidity, the initial grain moisture, and the feed rate. Collectively, it is often required for effective and robust control strategies in the challenging task. In this study, an advanced controller was presented to optimize the paddy circulation counter-flow drying, according to the theoretical framework of the model predictive control (MPC). The optimal configuration was specifically designed to accommodate the dynamic behavior and structural constraints. The controller was also used to dynamically regulate the paddy mass flow rate—the primary operational variable—in real time. Thereby, the residence time of the material was adjusted within the drying chamber. The entire drying trajectory was guided toward the final moisture content with high precision. A prediction model was embedded to forecast the future system states over a finite receding horizon. A sequence of the optimal control was computed to minimize the deviations from the target setpoints, according to the physical, operational, and actuator constraints. The objective function of the controller was formulated to incorporate an analytically derived attenuation coefficient, denoted as ‘β’. The penalty term was also applied, as the control input changed. As such, these modification was effectively relaxed the constraint on the paddy flow rate. More responsive adjustments were realized during transient conditions. The closed-loop stability was maintained to prevent excessive wear on the actuators or mechanical components. Simulations and experiments were conducted to evaluate the performance of the controller, including its robustness, resilience against unmeasured disturbances, and the accuracy of time-varying setpoints tracking. Specific case studies were designed to investigate the generalization of the controller for the precise control under abnormal operating scenarios, such as sudden equipment malfunctions. Thereby, the controller was then validated as suitable for the real-world industrial applications. Experimental results demonstrate that the MPC controller significantly mitigated the adverse effects of the major practical disturbances, particularly on some variations in the initial moisture content of the incoming wet paddy and ambient relative humidity. Among all test conditions, the maximum absolute deviation between the actual and target moisture content of the discharged paddy remained below 0.35% on a wet basis (wet basis), indicating the exceptional control accuracy and consistency. The controller was achieved remarkably in the low relative average deviations (RAD) of 0.11%, 0.07%, and 0.01%, respectively, under dynamic setpoint transitions, including the step changes, linear ramp profiles, and sinusoidal reference signals. The superior tracking performance was obtained to fully meet the varying operations. In the anti-interference tests, the artificial disturbances were introduced with the peak amplitudes of ±40%, ±60%, and ±90% at the outlet stage of the drying, in order to simulate the severe process upsets; The average relative deviation in the final moisture content was maintained at only 0.32%, 0.40%, and 0.41%, respectively, under these extreme conditions. Comparative analysis revealed that there was the decrease in the average relative deviation (RAD) between the actual and target moisture content of the paddy: The moisture content RAD decreased by 10.7%, 18.4%, and 30.5%, respectively, compared with the uncontrolled condition; The RAD decreased by 5.9%, 11.1%, and 21.2%, respectively, compared with the conventional PID control; The moisture content RAD decreased by 3.0%, 7.5%, and 14.6%, respectively, compared with the feedforward PID. Collectively, the outstanding performance of the controller was highlighted to reduce the impact of the unknown, variable, and potentially destructive disturbances caused by the abnormal or unpredictable operating conditions. Therefore, the optimal parameter can also provide for a theoretically sound and practically implementable framework for the process stability in the rice circulation.