Original scientific paper
ADAPTIVE CONTROL BASED ON THE APPLICATION OF A SIMPLIFIED UNIFORM STRUCTURES AND LEARNING PROCEDURES
Jozsef K. Tar
; John von Neumann Faculty of Informatics, Budapest Polytechnic, Budapest, Hungary
Miklos Ronto
; John von Neumann Faculty of Informatics, Budapest Polytechnic, Budapest, Hungary
Abstract
The present state of creating a new branch of Soft Computing (SC) for particular problem classes, possibly wider than the control of mechanical systems, is reported in this article. Like "traditional" SC il evades the development of analytical system models, and uses uniform structures, but these structures originate from various Lie groups. The advantages are a drastic reduction in size and an increase in lucidity. The generally "stochastic or semistochastic" "learning" or parameter tuning seems to be replaceable by simple explicit algebraic procedures of limited steps, too. The idea originated from mechanical systems' control while considering their general internal symmetry group, and later it was further developed by using specific general features of it on a much wider scale. Convergence considerations are given for MIMO and SISO systems, too. Simulation examples are presented for the control of the inverted pendulum with the use of the Generalized Lorentzian Matrices. It is concluded that the me/hod is promising and probably imposes acceptable convergence requirements in many cases.
Keywords
adaptive control; Soft Computing; Lie groups; partial system-identification; Lorentz group
Hrčak ID:
78701
URI
Publication date:
14.12.2000.
Visits: 1.017 *