Design of Robust Autoland Control Laws using mu-Synthesis

Abstract

The design of autoland control laws for a large civil transport aircraft using mu-synthesis is discussed. The controller architecture consists of functions for tracking the ILS-approach path, and for flare and alignment of the aircraft shortly before touchdown. For each function, design requirements are formulated using weighting functions in the frequency domain. Model uncertainty is incorporated via automatically generated Linear Fraction Parametric Uncertainty Models (LF-PUMs). The structured singular value, as the measure for robust performance, is then optimized via so-called D-K iteration. Robustness of the resulting controller functions is assessed via mu-analysis and batch simulations. Monte Carlo analysis is used to test autoland performance under varying environment and disturbance parameters. As an ultimate robustness test, this analysis is performed with the nominal, as well as worst-case aircraft model configurations. The reason for this is, that in practice the Monte Carlo aircraft model may be more detailed and accurate than the model available during the design, frequently requiring late additional design iterations.