Stochastic Model Predictive Control With Closed-Loop Model Updating

Abstract

Updating the process model remains an important concern in practical implementations of Model Predictive Control (MPC). This work introduces a novel stochastic model predictive control (MPC) framework with closed-loop model updating based on closed-loop operating data. The proposed framework is tested on a canonical chemical process case study. Simulations demonstrate that the economic performance is similar to a conventional MPC, but the model prediction accuracy and controller error are substantially reduced under all disturbance and operational scenarios considered.