Research of the MIMO structure and predictive control algorithms based on a digital twin of a distillation column

This paper examines the control of a distillation column as a multi-input-multi-output (MIMO) system with pronounced cross-couplings between process variables. Control channel interactions are analyzed using the relative gain array (RGA) method. A computational framework is developed based on a digital twin implemented as a MESH model and statistical MIMO-ARX models identified using synthetic data. A computational experiment is conducted comparing decoupled PI control, linear MPC, and nonlinear NMPC algorithms under identical conditions. The influence of cross-couplings and disturbances on the closed-loop system dynamics and the accuracy of maintaining process parameters is studied. The influence of the model structure and control method on the values of integral quality criteria and energy indicators is established. A reproducible computational approach based on the use of synthetic data for the analysis, identification, and testing of control algorithms is proposed.

Hali tarjima qilinmagan