Modular robust model predictive control

Abstract

This paper presents a method of combining model predictive control (MPC) with explicitly defined stability criteria for improved robust performance. The stability criteria is fundamental in its basis and can be applied universally to a process of any size. The stability criteria is independent variables based and can be used with linear or non-linear process; when used with linear process it imparts a quasi-linear optimal closed loop behavior. The stability criteria determines speed of optimization. A braking action can be included in conjunction with the stability criteria to permit a complete cycle control involving startup, normal operation and shutdown. The stability criteria supports both hierarchical and distributed MPC implementation consistently; that allows for the formation of a hierarchical and distributed MPC system within a process while permitting it to be connected to neighboring processes as part of a unified control system for an entire production chain involving a network of modular robust MPCs. Intrinsically, the stability criteria makes a MPC both robust and modular.