Multivariable Design Techniques Research Articles

A Study of Robust Multivariable Control Designs for Remotely Operated Vehicle Depth Control Systems

This paper looks at the application of five robust multivariable design techniques applied to the depth control of a nonlinear Remotely Operated Vehicle (ROV) subjected to parameter perturbations. The control problem is posed as a multi-objective optimization problem and it is shown to offer the designer an efficient, effective and intuitive approach for selecting the free control parameters for each of the controllers.

ON THE DESIGN OF ROBUST CONTROL SYSTEMS FOR DISTILLATION COLUMNS

Abstract The distillation column dynamics is very nonlinear, especially at high purity. Multivariable control system designs, which are essentially linear, may not be able to perform well at different operating conditions. This paper looks at three different multivariable design techniques—decoupling control, optimal state feedback and pole assignment as applied to distillation column control. Robustness of these techniques are analyzed by looking at the performance of these controllers at different operating conditions. Some interesting results are obtained.

The design of strip shape control systems for Sendzimir mills

A multivariable design technique is described for use in the design of strip shape control systems. The main control problem stems from the uncertainty in the knowledge of the mill gain matrix. The effect of modeling errors is to introduce interaction into the imperfectly compensated system. The design technique involves squaring down the system to make it more robust in the presence of such errors. The approximately upper triangular form of the transformed system enables a simple precompensator to be designed and the system is shown to be stable over a wide range of line speeds.

Some Considerations of Feedback Loop Design for the SRS Storage Ring R.F. System

A stability analysis has been made of the electron beam/r.f. cavity system for a high current storage ring under closed loop feedback conditions, using a variety of control theory techniques. A linearized, state variable model is formulated in terms of small changes in amplitude and phase of r.f. cavity voltage and generator drive. Modern multivariable design techniques are applied to the state space model. It is found that a multivariable root-locus method is more suited to this particular problem than is the inverse Nyquist array technique. Several cases are considered for both accumulation and storage modes of operation, with and without reactive mismatch between r.f. generator drive current and effective cavity volts. Amplitude and phase responses are obtained corresponding to step changes and several possible feedback options, involving proportional or integral action, are presented.

Diagonal dominance and the method of pseudodiagonalisation

Very little practical guidance has been forthcoming on achieving a diagonally-dominant system prior to controller design, using either the Nyquist or inverse-Nyquist multivariable design techniques. In this reappraisal of pseudodiagonalisation and diagonal dominance, a linear space setting is used as a framework for the existing technique of Rosenbrock and several new related optimisations. The relationship between pseudodiagonalisation and dominance at one given frequency is discussed. In particular, sufficient conditions, for successive iterates of a pseudodiagonalisation procedure, to create dominance at that frequency are given. The extension of pseudodiagonalisation to try and create diagonal dominance per se is examined in some detail. The new technique for pseudodiagonalisation is utilised in a least-squares approach to diagonal dominance, and the new algorithm demonstrated on the example of Hawkins.