Structural design for on‐line process optimization: I. Dynamic economics of MPC

Abstract

AbstractThe structural design of integrated online process optimization and regulatory control systems based on an economic analysis of different structures is addressed. The regulatory control layer is assumed to be implemented using model predictive control (MPC) techniques. An approach to the analysis of the dynamic economics of MPC is presented which uses the state‐space formulation as the plant model. Output feedback is performed in the framework of linear quadratic filtering theory using a Kalman filter. Using the unconstrained model predictive control law, the variance of the constrained variables of the closed‐loop system subject to stochastic disturbances is analyzed. Based on the variance of the constrained variables, the amount of necessary backoff from the constraints due to regulatory disturbances is calculated and the dynamic economics are established. The dynamic economics of the model predictive regulatory control system are incorporated into the method of the average deviation from optimum analyzing the economic performance of an online optimization system. Thus, different structures of the integrated system of online optimization and MPC‐based regulatory control can be analyzed in terms of their economic performance, and the necessary structural design decisions can be taken.