Adaptive Closed-Loop Control System for the Optimization of Tablet Manufacturing Processes

Background: Tablet manufacturing is challenged by strong dynamic coupling of process parameters, significant material property fluctuations, and delayed quality control, with tablet weight stability being particularly critical in high-speed production. Traditional static optimization methods relying on empirical judgment struggle to manage these dynamics, leading to substantial variations in tablet weight and hardness that severely compromise production efficiency. Methods: This study proposes a data-driven closed-loop control system centered on a novel Iterative Learning Model Predictive Control (IL-MPC) architecture. The core innovation lies in directly integrating iterative learning constraints within the MPC optimization framework. This constraint-embedding mechanism enables systematic utilization of historical batch data while preserving real-time optimization capabilities. The IL-MPC approach achieves enhanced batch-to-batch performance consistency with reduced computational burden, effectively combining the dual advantages of learning and optimization. Results: Simulation experiments and industrial production data validate the practical feasibility of the IL-MPC algorithm. Implementation results demonstrate that the proposed system effectively manages dynamic process variations, significantly improving control precision for both tablet weight and hardness, outperforming conventional control methods. Conclusions: This research breaks through the technical bottleneck of dynamic regulation in tablet manufacturing. The developed IL-MPC framework provides a reproducible closed-loop control paradigm for intelligent pharmaceutical manufacturing, promoting the industry's transformation toward data-driven models and advancing intelligent drug production.