This paper considers a model-following controller with optimal disturbance attenuation. The controller design process follows a similar approach to the MDM/MDP (Multiple Delay Models/Multiple Design Points) design method, in which the controller performance is evaluated only at selected design points and optimization is performed numerically. Unlike the MDM/MDP method, however, performance is evaluated over the whole range of uncertainties, and is optimized with a method which has a convergence property. In this design process, the performance of the disturbance attenuation is defined as a quadratic cost function and the design problem is formulated with one convex and one non-convex condition, but this is not easily solvable. A design procedure that uses an iterative algorithm, which is easily solvable and has the convergence property, is presented to render the problem tractable. Next, model-following controllers for the lateral/directional motions of MuPAL-α, one of the National Aerospace Laboratory’s flight experiment aircraft are designed with the proposed method, and the results of flight experiments to verify their performances are presented.
抄録全体を表示