Model predictive control of high moisture extrusion cooking

High Moisture Extrusion Cooking (HMEC) has become a promising technology for producing plant-based meat alternatives. By using HMEC, food manufacturers can create meat-like textures from plant proteins, offering a sustainable solution with reduced carbon footprint to consumers. However, at the current stage of development, the automation level in HMEC is insufficient to ensure operational autonomy, reliability, and product quality expected by industry demands. This paper presents a predictive control framework designed to transform experience-based handled HMEC into a more reliable process operation, improving its production performance and facilitating industrial up-scaling. The proposed control structure is hierarchical, comprising two layers. At the upper layer, a model predictive control (MPC) algorithm determines the optimal set-points for the controllers at the lower layer. The predictive framework is built on the existing HMEC control architecture and can be further extended to achieve fully optimized production. Leveraging linear dynamic models, the approach mainly focuses on the protein melt control aiming to enhance production performance by minimizing the tracking error of process quantities correlated to product quality. The practical feasibility of the designed control solution has been proven on a pilot-scale extruder. Validation results have shown improved operational stability and reproducibility, while effectively tracking set-points for consistent meat-like fibrous structure formation and desired textural characteristics.