F. Crusca

Design and implementation of a gain scheduling controller for a water level control system.

The design of a gain-scheduled proportional plus integral (PI) controller for a nonlinear water-tank level control system is presented. A novel approach of determining operating points in terms of interpolation error is also given. The controller is implemented in LABVIEW in the laboratory. The field test results of the system response at different operating conditions clearly illustrate the controllers improved performance over a fixed PI controller.

Multivariable frequency-domain techniques for the systematic design of stabilizers for large-scale power systems

A novel application of multivariable frequency-domain control theory to the design of excitation based stabilizers for large-scale electric power systems is presented. The stabilizers are designed to coordinate the global performance of the multimachine system, and in some cases may turn out to be of a decentralized structure. The design procedure is based on the direct Nyquist array, supplemented by the characteristic function. The control design procedure is applied to a sample twelve-machine electric power system which includes detailed models of synchronous machines, excitation systems, turbines, and speed governing mechanisms. Simulation results show that the design method leads to the removal of the underdamped oscillations, to a reduction in the interaction between the generating units, and to a significant improvement in the dynamic performance of the system. This is the first successful application of the direct Nyquist array method to the systematic design of excitation-based power system stabilizers for large-scale power systems. >

A Comparison of Multivariable Time- And Frequency-Domain Design Methods for Power System Stabilisers including Links with Reported Field Trials

The paper describes a comparison of multivariable, frequency-domain and time-domain design methods to derive supplementary stabilising signals for a highly oscillatory electrical power system. A direct comparison of different multivariable design approaches is effected on the same power system for the first time. The direct Nyquist array (DNA) method is extended in its application to produce practical stabiliser structures not previously obtained by other multivariable design approaches. All of the controllers designed have been implemented in practice including the DNA extension which produces a stabiliser requiring the measurement of machine terminal quantities only. The different theoretical approaches are also shown to produce essentially the same stabilisers and this, combined with their practical realisability, is a unifying result which is both satisfying and worthwhile emphasising.

Mixed sensitivity H/sub infinity / controller design for power systems

An H/sub infinity /-based controller is designed for a large-scale electrical power system to improve its dynamical performance. Emphasis is placed on reducing the interaction among generator terminal voltage outputs. By means of the mixed sensitivity approach, a significant improvement in the sensitivity gain over the design frequency range was obtained, while maintaining a satisfactory margin of robustness against output multiplicative model perturbation. The results obtained indicate the potential applicability of H/sub infinity / theory to the design of controllers for power systems.<<ETX>>

Quadratic cost function design for linear optimal control systems

A systematic method for choosing a quadratic cost function for linear optimal control systems is presented. The method is based on the identification of the most controllable and observable modes of the system. The method is used for the design of an optimal proportional-plus-integral controller for a large-scale electric power system comprising 12 machines, transmission lines and load centers. It is shown that the controller obtained by the proposed method significantly improves the dynamic performance of the power system.<<ETX>>

A robust least square fault detection approach for linear systems with structured time-varying perturbations

This paper proposes a robust least square approach to the robust fault detection and estimation problem for a linear time-invariant (LTI) system. Rather than using a reference model and transfer the fault detection problem into a model-watching formulation, we provide a direct fault reconstruction approach to estimate the incoming fault signal. Unknown inputs, actuator faults, sensor faults and disturbances are considered in our design. A sufficient affine matrix inequality condition is developed which guarantees that the filter estimation error is kept below a specified level of performance index in the presence of disturbance and structured time-varying norm-bounded uncertainties. In order to transfer the nonlinear matrix inequality into a equivalent affine problem, we define certain congruence transformation parameters and provide an interior-point convex optimization method to obtain the convex solution. A numerical example is provided to confirm the effectiveness of our approach.

Modular Modelling of Nonlinear Complex Systems

A new modular approach to the modelling of nonlinear complex systems is introduced in this paper. The approach draws on the graphical capabilities of the combined Simulink© and Matlab© software packages. In this approach, a complex system is decomposed into self-contained modules with well defined sets of inputs and outputs, so as to facilitate the easy linking of individual models. A power system comprising a synchronous machine, an exciter, and a network of buses, lines and loads is used as example to demonstrate the modelling approach. Simulation results of the system at certain specified operating conditions are presented.

A computer aided learning package using the World Wide Web

The World Wide Web can be used a powerful and convenient means of disseminating computer aided education because most students already have access to the Web and the necessary Web browsers such as Netscape. There are many benefits in this approach because students already tend to do significant amounts of Internet surfing and are familiar with the usage of Web sites and browsers. Therefore online instruction for their coursework using this all pervasive computer network which may be accessed from almost anywhere is a powerful and beneficial low cost alternative to other forms of computer aided learning. The key aims of this package development are to provide a compact and self contained package of instruction in the features and facilities provided by the C programming language together with an introduction to the Unix programming environment with its vast range of utilities and command interpreter shells. Part of this introduction to Unix is provided in the Bourne Shell as a programming language instruction module.

Toolkit for large-scale systems controller design in the frequency-domain

Abstract A novel software package specifically structured for large-scale control systems is presented in this brief paper. A large-scale system comprising 99 states, 10 inputs and 10 outputs is taken as an example to demonstrate the improtant features of the package, through a controller design exercise.

Programs for CAD of Multivariable Control Systems in both the Time and Frequency Domains

Abstract In this paper a description of a suite of programs used in the computer-aided design of linear, timeinvariant, multivariable control systems is given. Algorithms which have been programmed for control systems design in the time-domain include optimal state feedback, Kalman filtering, pole-placement, observer design, and output feedback design with dynamics if necessary. The option to design controllers for continuous- or discrete-time systems is generally provided in the above programs. Controller design in the frequency-domain is carried out using a package which incorporates the direct-Nyquist array, inverse-Nyquist array, and characteristic locus design methods. User interaction with this package is via a vector-refresh graphics workstation. The foregoing controller design algorithms in turn rely upon a number of support programs. These may include eigenvalue and eigenvector routines, function minimisation routines and LU Factorisation routines, for example. Programs for the conversion between time- and frequency-domain system representations are also available. Associated programs for the manipulation of polynomials include a root-finding routine, and for matrix polynomials a greatest common left- or right-divisor routine, and the Smith-normal form routine. A routine for finding the McMillan form of rational matrices is also available. Other support facilities include a matrix utility which allows the storage and retrieval of system data in a common database- as well as the execution of standard matrix operations -- and a plot facility for the time-simulation of open- or closed-loop systems. The entire suite of design, simulation and support programs described above is supported by a user-friendly, help facility. Examples of interactive design sessions, including the design of a power system stabiliser, are given in the paper, with results being taken directly from computer hardcopy.