Search

This article extends two recent contributions in the field of quantitative feedback theory to the multivariable case. They concern the model matching and the measured disturbance rejection problems. The model matching problem is a tracking control problem with specifications given as acceptable deviations from an ideal response. The measured disturbance rejection problem balances feedback and feedforward actions to reject disturbances. Both perspectives present advantages over classical quantitative feedback theory techniques in certain situations. This paper develops the necessary tools to solve both control problems in the case of multi-input multi-output plants. In particular, it shows how to derive nonconservative controller bounds for each of the single-input single-output control problems in which the overall multivariable problem is divided. The result is a systematic controller design methodology for multi-input multi-output plants with structured uncertainty. The application of the technique to the well-known quadruple-tank process illustrates the benefits of the method. Copyright © 2016 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

SUMMARY This paper introduces a novel solution for the multi-input multi-output (MIMO) quantitative feedback theory control design problem with tracking error specifications. Looking for a minimum controller overdesign, the technique finds new controller quantitative feedback theory bounds based on necessary and sufficient conditions for the existence of suitable associated prefilter matrix elements. It improves previous approaches to the subject and includes (i) the possibility of a free selection of the nominal plant, (ii) a less conservative application of the Schwartz inequality to decisively reduce the potential controller overdesign, (iii) a methodology to design independently the elements of the prefilter matrix, and (iv) a scope of application to both sequential and nonsequential MIMO controller design methods. The benefits of the new control design technique are illustrated by means of two examples. The first one, a standard 2 × 2 MIMO problem, is provided for comparison purposes with previous approaches. The second example, included as a major control challenge, deals with a well-known demanding distillation column benchmark problem. Copyright © 2013 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]

SUMMARY An alternative to the traditional QFT tracking problem, in which upper and lower tolerances are imposed on the magnitude of the tracking transfer function, is to define the robust specification as a boundary on the deviation of such function from a predefined model. However, previous research exploring this approach reveals a certain overdesign and dependence on the choice of the nominal plant. This paper establishes a necessary condition on the controller from tracking error specifications. With this condition, the controller bounds introduce no overdesign, and the resulting two-degrees-of-freedom design is independent of the choice of the nominal plant, similar to the traditional approach. Copyright © 2011 John Wiley & Sons, Ltd. [ABSTRACT FROM AUTHOR]