Parameter Varying Control of a High-Performance Aircraft

Abstract

Thedynamicresponsecharacteristicsofmodern aircraftvary substantially withe ightconditions.Thesechanges require scheduling of the e ight control system with variables such as dynamic pressure and Mach number. This scheduling can be accomplished easily for simple controllers but is much more dife cult for complex controllers, which result from the use of most modern control design techniques. On the other hand, these complex controllers can yield signie cant performance improvements when compared with simple controllers. Recently, linear parameter varying (LPV) techniques have been developed that provide a natural method for scheduling H1 based controllers. LPV techniques are combined with π synthesis methods to develop a self-scheduled longitudinal controller for a high-performance aircraft. The ability of this controller to achieve specie ed handling qualities over a wide range of e ight conditions is demonstrated by nonlinear simulations. I. Introduction M ODERNhigh-performanceaircraftoperateoverawiderange of e ight conditions. This results in dynamic response characteristics that vary substantially during a typical mission. Traditionally, e ight control systems were designed by using mathematical models of the aircraft linearized at various e ight conditions. Relatively simple e xed-structure control laws were formulated and gainswereselectedforeache ightconditionbyusingclassic,singleinput, single-output methods. Because the structure of these control laws was simple, only a few gains needed to be scheduled and, therefore, scheduling was fairly easy. As aircraft have become more complex with a variety of control effectors and sensors and as performance capabilities and requirements have increased, traditional methods for controller design often have not yielded acceptable performance. Thus, the use of various modern multiinput, multioutput techniques for e ight controller design has been extensively studied. These techniques use linearized models of the aircraft dynamics but result in controllers that are much more complex and that use many more gains than those designed by classic methods. Consequently,itismuchmoredife culttoschedulecontrollersdesigned by modern techniques,andthishasbeen amajorimpediment in the use of these theories in the design of e ight control systems. Recently, a number of investigators have proposed the use of dynamic inversion together with π synthesis methods for the design of aircraft e ight control systems. 1i4 Dynamic inversion avoids the scheduling problem by using feedback to cancel the dynamics of the aircraft. Desired dynamics are then substituted for the canceled dynamics. Some promising preliminary results have been obtained by using dynamic inversion but there are some important implementation issues that may inhibit the use of this method in practice. In this paper, we describe the application of an extended H1 technique to the design of a self-scheduled controller for the longitudinal control of a high-performance aircraft. This technique is based on linear parameter varying (LPV) techniques, which result in a controller that is scheduled with dynamic pressure. The closedloop, LPV control structure can be represented as shown in Fig. 1.