Reduced Order Nonlinear Multi-parametric Model Predictive Control of Large Scale Systems

Multiparametric MPC (mp-MPC) is an efficient methodology for computing fast control actions. Control of distributed parameter systems (DPS) remains a challenging task, as the system dynamics are infinite-dimensional. Model reduction of such systems may produce instabilities and thus it is essential that the model reduction methodology used is robust. In this work, a Galerkin-based, efficient model reduction is employed for DPS. The reduced order model is then utilized by the mp-MPC, introducing a novel strategy for computing the critical regions of the approximated multiparametric nonlinear problem. This strategy initializes the solution effectively, reducing the number of the required critical regions and the corresponding computational time. The effectiveness of the proposed algorithm is demonstrated for a tubular reactor where an exothermic reaction takes place