Design of uav robust autopilot based on adaptive neuro-fuzzy inference system

• Main Authors: Mohand Achour Touat, Anatoly A. Tunik
• Format: Article
• Language: English
• Published: National Aviation University 2008-04-01
• Series: Vìsnik Nacìonalʹnogo Avìacìjnogo Unìversitetu
• Online Access: http://jrnl.nau.edu.ua/index.php/visnik/article/view/1624

<p><em> This paper is devoted to the application of adaptive neuro-fuzzy inference systems to the robust control of the UAV longitudinal motion. The </em><em>adaptive neore-fuzzy inference system</em><em> model needs to be trained by input/output data. This data were obtained from the modeling of a ”crisp” robust control system. The synthesis of this system is based </em><em>on the separation theorem, which defines the structure and parameters of LQG-optimal controller, and further - robust optimization of this controller, based on the genetic algorithm. Such design procedure can define the rule base and parameters of fuzzyfication and defuzzyfication algorithms of the </em><em>adaptive neore-fuzzy inference system</em><em> </em><em>controller, which ensure the robust properties of the control system. Simulation of the closed loop control system of UAV longitudinal motion with </em><em>adaptive neore-fuzzy inference system</em><em> </em><em>controller demonstrates high efficiency of proposed design procedure. </em></p>