抄録
This paper considers an adaptive flight control system using neural networks. In this system, neural networks are aimed to obtain an inverse dynamics of an aircraft online and the feedback error learning (FEL) strategy proposed by Kawato et al is used as their learning scheme. In FEL, neural networks are arranged parallel to a conventional feedback (CFB) controller as a feedforward (FF) controller which makes it possible to cover the shortcomings of the CFB controller and improve the whole control performance. Numerical examples demonstrate the control process of an aircraft under various situations and show that the constructed control system is able to improve the control performance by learning the local inverse dynamics of an aircraft.
