A Robust Non-linear Design for Autopilot Control Law Using Dynamic Inversion and H-infinity Loop-shaping

Abstract. This paper presents a robust control law design for a commercial aircraft pitch autopilot dedicated to perform the approach maneuver. The design involves only the longitudinal axis. Traditionally, in the aerospace industry, the design of autopilot functions follows a process that involves the partitioning of the flight operational envelope into separate flight conditions, the design of a compensator, for each of the flight conditions, to satisfy closed-loop specifications and, finally, a gain schedule is designed to make the final control law cover the full flight envelope. However, the practice shows that the resulting control law is not able to deal satisfactorily with plant variations and the intrinsic non-linear behavior of the aircraft. The process outlined in this paper combines two modern control law design techniques, namely Dynamic Inversion (DI) and H-infinity loop-shaping, in order to provide a robust control law. The design is divided into two stages: (a) Dynamic Inversion is used in the inner-loop design to generalize the dynamic of the aircraft. It uses feedback linearization in order to deal with the aircraft nonlinear behavior and replaces it by a linear time invariant system defined by the inner-loop compensator, witch eliminate the design of a gain schedule. (b) Robust H-infinity loop-shaping technique is used on the outer-loop design. The objective is to adjust the open-loop frequency response of the system to a desirable reference system with specified robust stability and performance requirements. The designed compensator adjusts both the open-loop singular value frequency response to the reference system and a coprime factor that is directly related to the stability and performance margins. The final result is evaluated by means of frequency domain singular value response and time domain simulation, using Dryden wind disturbance mode.