The modular multivariable controller: I: Steady-state properties
Abstract
The modular multivariable controller (MMC) represents a multivariable controller design methodology which is based on the solution of multiobjective optimization problems using the strategy of lexicographic goal programming; priority-driven, sequential satisfaction of objectives. This article formally introduces the concept of the MMC, analyzes its static characterstics, and proposes a specific methodology for the design of steady-state MMCs. It is shown that the framework of MMC can explicitly handle all types of control objectives (for example, equality or inequality specifications on controlled outputs), and constraints on manipulations. Its priority-driven, sequential satisfaction of control objectives leads to a modular, hierarchical structure of controllers with specific objectives. The modular character of MMC allows the explicit maintenance, tuning, and reconfiguration of multivariable control systems, while its hierarchical structure explicitly expresses engineering decisions and trade-offs. Its static design incorporates uncertainty in process gains and automatic reconfiguration to account for failure in sensors and/or actuators. The design of an MMC for a heavy oil fractionator is presented to illustrate the controller's character and the proposed methodology for the design of static MMCs.