A synthesis of optimal neurocontroller ensemble for multiple-regime aircraft

Аuthors

Brusov V. S.^*, Tiumentsev Y. V.^**

Moscow Aviation Institute (National Research University), 4, Volokolamskoe shosse, Moscow, А-80, GSP-3, 125993, Russia

*e-mail: vsbrusow@mail.ru
**e-mail: yutium@gmail.com

Abstract

Application possibilities are discussed for artificial neural networks to solve problems associated with aircraft motion control. Aircraft are treated as multiple-regime dynamical systems (MDS). It is suggested to use neural controller ensemble (NCE) for a case of essentially multiple-regime systems. According to this approach we begin with a synthesis of a model for controlled MDS, then we build a neural controller for a single-regime system and generate an index to evaluate efficiency of the controller. After that a concept of the NCE is introduced based on the results for the single-regime system controller as well as an efficiency index is derived for the MDS case. Three essential optimization problems are formulated which are typical for the NCE assigned for MDS. Main features of the considered approach are illustrated via solving of a control problem for a simple MDS.