A study on switched linear system identification using game-theoretic strategies and neural computing

Abstract

This study deals with application of game-theoretic strategies and neural computing to switched linear system identification, wherein some of the subsystems may be in failed, standby, or working states. The controller is to detect failed subsystems, and switch standby and working subsystems to maintain stable system configuration. Its performance is based on a number of successful stabilizing controls before entire system failure. A strategy board game is redefined as switched linear system and used in the experimental study. Mathematical model of the system is derived from its physical description, and the controller is modelled as adaptive neural network. The analysis results to estimation of system parameters and identification of stabilizing switching control rules. The system trajectory is ascertained by simulation tests, wherein subsystem failure time intervals are varied. The study demonstrates feasibility of the non-classical methods in system identification from first principles. Besides prototype system identification, our methodology has applications in combinatorial switching control modelling and experimental validation of mathematical models of switched linear systems.

Journal of the Nigerian Association of Mathematical Physics, Volume 19 (November, 2011), pp 61 – 68