MIMO model of an interacting series process for Robust MPC via System Identification

Area of Science:

• Process Control
• System Identification
• Empirical Modeling

Background:

• Interacting series processes present complex dynamics challenging for accurate modeling.
• Pilot plant experiments are crucial for validating process identification techniques.
• Model predictive control (MPC) requires precise process models for effective implementation.

Purpose of the Study:

• To investigate empirical modeling of an interacting series process using system identification.
• To evaluate the performance of different system identification approaches.
• To assess the real-time implementation of identified models in a linear model predictive control system.

Main Methods:

• Experimental study using a gaseous pilot plant.
• Application of system identification techniques for process modeling.
• Real-time implementation and evaluation of linear model predictive control.

Main Results:

• Three practical system identification approaches were investigated and compared.
• The selected empirical model successfully reproduced open-loop plant dynamics.
• The identified model achieved zero steady-state errors in closed-loop control.

Conclusions:

• System identification provides effective empirical models for interacting series processes.
• The developed models are suitable for real-time control applications, including MPC.
• Addressing MIMO state-space model construction is key for series interacting processes.