ROBUST AUTOPILOT DESIGN STRATEGY FOR FUNCTIONALLY DEPENDENT PARAMETERUNCERTAINTIES

In this paper an approach is presented to incorporate known functional dependencies of model parameters on uncertain physical quantities. Th e approach is suitable in cases where the source of the plant uncertai nty is mainly accountable by a few unknown physical parameters. In suc h cases, uncertainty over-bounding is greatly reduced compared with us ing general uncertainty descriptions. This may lead to a better perfor mance/robustness trade-off in a robust controller design. The basis of the approach is to describe the model uncertainties in the unknown ph ysical parameters as linear fractional transformations (LFTs). Then, u sing LFT algebra, an LFT is obtained, which describes the effect of th e uncertain variables on the plant model. The resulting model can be u sed with several robust control synthesis techniques. In particular, i n this paper le synthesis, which is most suitable for the resulting fu nctional dependence model was adopted. As an example, an aircraft cont roller to follow roil angle commands while keeping the lateral acceler ation close to zero was designed. The resulting controller satisfies t he specified performance requirements and its superior characteristics are demonstrated by comparison with a controller which was designed u sing a general uncertainty bounds model. (C) 1998 John Wiley & Sons, L td.