Identification and design of control systems

• Main Author: Yu, Cheng-Kung
• Published: University of Warwick 1985
• Subjects: 629.8; TJ Mechanical engineering and machinery
• Online Access: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.293862

An extensive literature on parameter identification and design of multi-input multi-output control systems exists. Despite this and the presence of a wide ranging software in these areas of control engineering, there is an absence of analytical work and algorithms which are able to treat continuous linear and non-linear as well as discrete systems within the same mathematical framework. Furthermore, virtually all currently available identification and design algorithms require large processing power/working storage and, consequently are suitable only to users with access to mainframe computers. This thesis is concerned with the development of a number of novel mathematical theories for parameter identification and design of linear and non-linear systems. These theories and their associated results are used to develop numerical algorithms suitable for implementation on low-cost personal computers with memory sizes of around 256 kbytes. The programs described in the thesis have been written by using IBM—4331 Fortran and can be run with little modification on IBM—PC Although there are constraints on accuracy, the analytical results and the associated software developed in this thesis would be of use to practising control engineers in the preliminary analysis/ design of physical systems. The thesis is divided into two parts : Part I ( Chapters 1—5 ) considers the problems of parameter identification of continuous linear time - invariant as well as non-linear and time-varying and discrete systems. Part II (Chapters 6 and 7) considers the problems of designing digital servo systems. The main contributions of this thesis are: 1. Development of a new algorithm for the identification of the parameters of multi-input multi-output known order linear systems. This method is based on the integration of the completely controllable dynamical equation (Chapter 2). 2. Formulation of a novel method to identify the system order and the unknown parameters of multiple-input multiple-output linear time-invariant continuous systems. This method is based on the special structure of the system matrix and multiple integration of the dynamical equation C x(t)=Ax(t) + Bji(t) and ^(t)=C x(t) ) (Chapter 3). In Chapter 4, these results are extended to time-varying and non-linear systems. 3. Identification of system parameters and order for linear discrete single-input single-output systems using the rank difference between two matrices constructed from the input- output measurement (Chapter 5) . The procedure used to identify unknown parameters is based on a special output data vector y (K ; N) which is a linear combination of the input -output data The originality of the results presented in Chapters 2 — 5 lies in the fact that linear and non-linear as well as continuous and discrete systems may be identified by using a single mathematical framework. The main advantage of this is that a self-contained system package can be developed. Various parts of such a system software have been described in the thesis. 4. A microprocessor-based output deadbeat controller for digital servo systems with finite settling time is presented. The controller configuration introduced in this chapter can also be applied to implement the widely used digital lead-lag compensators and PID controllers (Chapter 6). 5. Anew method of designing digital systems using the parameter optimization method via non-linear programming and the finite-settling time criterion (Chapter 7).