Economic and Distributed Model Predictive Control of Nonlinear Systems
Jun 15, 2012
from 08:30 AM to 10:30 AM
|Where||ENGR. IV Bldg. Tesla Room 53-125|
|Add event to calendar||
Advisor: Professor Panagiotis D. Christofides
Maximizing profit has been and will always be the primary purpose of optimal process operation. Within process control, the economic optimization considerations of a plant are usually addressed via a two-layer architecture. In general, this architecture includes: the upper layer that optimizes process operation set-points taking into account economic considerations using steady-state process models, and the lower layer (i.e., process control layer) whose primary objective is to employ feedback control systems to force the process to track the set-points.
Optimizing closed-loop performance with respect to general economic considerations for nonlinear systems in a unified framework has recently become a subject of increasing theoretical interest with important practical ramifications. In addition to a tighter integration of economics and control, advances in communication technologies have motivated augmentation of traditional point-to-point and wired local process control systems with additional cheap and easy-to-install networked sensors and actuators and control systems. Such networked, distributed control systems can substantially improve the efficiency, flexibility, robustness and fault tolerance of an industrial control system while reducing installation, reconfiguration and maintenance expenses at the cost of coordination and design/redesign of the various control systems in the new architecture.
This dissertation presents rigorous, yet practical, methods for the design of economic and distributed model predictive control systems. Beginning with a review of recent results on the subject, the dissertation presents the design of Lyapunov-based economic model predictive control systems for a broad class of nonlinear systems using state and output feedback. Then, the dissertation focuses on the development of an economic model predictive control method with guaranteed improvement in closed-loop performance compared to conventional (i.e., incorporating a quadratic cost) Lyapunov-based model predictive control designs. Subsequently, the dissertation focuses on the design of a networked distributed model predictive control method for nonlinear uncertain systems subject to communication disruptions, multirate sampling and measurement noise, as well as of a distributed model predictive control method for nonlinear switched systems to compute optimal manipulated input trajectories that achieve desired stability, performance and robustness specifications. Throughout the thesis, the control methods are applied to nonlinear chemical process networks and their effectiveness and performance are evaluated through extensive computer simulations.
Mohsen Heidarinejad received his B.Sc. degree from the Sharif University of Technology, Tehran, Iran and his M.Sc. degree from the University of Toronto, Toronto, Canada in 2006 and 2008, respectively. Currently, he is a PhD candidate in Electrical Engineering Department at the University of California, Los Angeles (UCLA). His research interests include networked control systems, model predictive control and economic optimization.