Conference Paper

A hierarchical optimization method of large-scale time-delay systems

University of New Mexico, Albuquerque, NM
DOI: 10.1109/CDC.1981.269427 Conference: Decision and Control including the Symposium on Adaptive Processes, 1981 20th IEEE Conference on, Volume: 20
Source: IEEE Xplore

ABSTRACT In this paper a large-scale time-delay system is decomposed into N non-delay subsystems. A sub-optimal control based on heirarchical optimization of large-scale systems is then proposed and an algorithm for it is given.

  • [Show abstract] [Hide abstract]
    ABSTRACT: In this paper two hierarchical schemes are proposed for large-scale time-delay systems. The models of such systems are decomposed into a number of subsystems. Two fast-converging algorithms based on the extensions of interaction prediction and costate prediction methods to time-delay systems are presented. The algorithms are demonstrated by two examples.
    American Control Conference, 1984; 01/1984
  • [Show abstract] [Hide abstract]
    ABSTRACT: The paper is concerned with the problem of robust H<sub>∞ </sub> filtering for a class of systems with parametric uncertainties and unknown time delays under sampled measurements. The parameter uncertainties considered are real time-varying and norm-bounded, appearing in the state equation. An approach has been proposed for the designing of H<sub>∞</sub> filters, using sampled measurements, which would guarantee a prescribed H<sub>∞</sub> performance in the continuous-time context, irrespective of the parameter uncertainties and unknown time delays. Both cases of finite and infinite horizon filtering are studied. It has been shown that the above robust H<sub>∞</sub>-filtering problem can be solved in terms of differential Riccati inequalities with finite discrete jumps
    IEEE Transactions on Automatic Control 08/1998; · 3.17 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: The costate prediction method of hierarchical control of non-linear systems is based on two substitution relations for the first- and second-level problems in a multilevel formulation. The method has shown superior computational behavior for unretarded systems as compared to other schemes such as goal coordination or interaction prediction. In this paper the costate prediction is extended to retarded continuous-time and discrete-time systems. Algorithms are developed and illustrated for numerical examples.
    American Control Conference, 1982; 07/1982