12th AIAA International Space Planes and Hypersonic Systems and Technologies AIAA 2003-7081

15 - 19 December 2003, Norfolk, Virginia

Barı¸s Fidan†¶ , Maj Mirmirani‡# , and Petros A. Ioannou§¶

University of Southern California Los Angeles, CA 90089
California State University, Los Angeles, CA 90032

Abstract Nomenclature The following notation is used
throughout the paper, unless otherwise stated.
The current air-breathing hypersonic flight (AHF)
technology programs focus on development of flight a∞ : free stream velocity of sound
test vehicles and operational vehicle prototypes that A¯D : diffuser exit/inlet (area) ratio
utilize airframe-integrated scramjet engines. A key is- c¯ : reference length
sue in making AHF feasible and efficient is the con- CD : drag coefficient
trol design. The non-standard dynamic characteristics CL : lift coefficient
of air-breathing hypersonic flight vehicles (AHFVs) Cm : pitching moment coefficient (pmc)
together with the aerodynamic effects of hypersonic Cm (q) : pmc due to pitch rate
flight make the system modeling and controller design Cm (α) : pmc angle of attack
highly challenging. Moreover the wide range of speed Cm α : ∂Cm /∂α
during operation and the lack of a broad flight dynam- Cm (δe ): pmc due to δe
ics database add significant plant parameter variations CT : thrust coefficient
and uncertainties to the AHF modeling and control fs : stoichiometric ratio for hydrogen, 0.029
problem. In this paper, first, different approaches to h : vehicle altitude
this challenging problem presented in the literature are I (In ) : the (n × n) identity matrix
reviewed. Basic dynamic characteristics of AHFVs are Iyy : vehicle y-axis inertia per unit width
described and various mathematical models developed m : vehicle mass
for the flight dynamics of AHFVs are presented. Ma- m¯ : vehicle mass per unit width
jor nonlinearity and uncertainty sources in the AHF m ˙ air : air mass flow rate
dynamics are explained. The theoretical and practical m ˙f : fuel mass flow rate
AHF control designs in the literature, including the M : pitching moment
control schemes in use at NASA research centers, are M∞ : vehicle flight Mach Number
examined and compared. The review is supported by a nx : acceleration along the vehicle x-axis
brief history of the scramjet and AHF research in order nz : acceleration along the vehicle z-axis
to give a perspective of the AHF technology. Next, the P : pressure
existing gaps in AHF control and the emerging trends q : pitch rate
in the air-breathing hypersonic transportation are dis- Q : generalized elastic force
cussed. Potential control design directions to fill these re : radial distance from Earth’s center
gaps and meet the trends are addressed. The major Re : radius of the Earth, 20,903,500 ft
problem in AHF control is the handling of the vari- S : reference area
ous coupling effects, nonlinearities, uncertainties, and T0 : temperature across the combustor
plant parameter variations. As a potential solution, Th : thrust
the use of integrated robust (adaptive) nonlinear con- u : speed along the vehicle x-axis
trollers based on time varying decentralized/triangular V : vehicle velocity
models is proposed. This specific approach is moti- X : force along the vehicle x-axis
vated by the promise of novel techniques in control Z : force along the vehicle z-axis
theory developed in recent years. α : angle of attack
γ : flight path angle (γ = θ − α)
∗ This work was supported in parts by Air Force Office of Sci-
δe : pitch control surface deflection
entific Research under Grant #F49620-01-1-0489 and by NASA δt : throttle setting
under grant URC Grant #NCC4-158.
† Student Member AIAA, graduate student, Electrical Engi- ∆τ1 : fore-body elastic mode shape
neering Department. ∆τ2 : after-body elastic mode shape
‡ Member AIAA, professor, Mechanical Engineering Depart-
ζ1 : damping ratio of the first vibration mode
ment. η : generalized elastic coordinate
§ Professor, Electrical Engineering Department

American Institute of Aeronautics and Astronautics

Copyright © 2003 by the American Institute of Aeronautics and Astronautics, Inc. All rights reserved.

ηf : fuel equivalence ratio, fs m˙ fair engines.6 Scramjet technology is an enhanced version
θ : pitch angle of the ramjet technology, in which the engines oper-
µg : gravitational constant ate by subsonic combustion of fuel in an air stream
ρ : density of air compressed by the aircraft’s forward speed. In con-
ω1 : natural frequency of the first vibration mode ventional turbojet engines, the compressor fans com-
0n×m : the n × m zero matrix press the air. Scramjets are basically ramjet engines
in which the airflow through the engine is kept su-
personic. Using hydrogen as the basic fuel for com-
Subscripts bustion, the air-breathing scramjet engines burn oxy-
A : due to aerodynamics
gen scooped from the atmosphere. Obtaining the
E : due to external nozzle
oxygen from air rather than carrying it on-board,
T : due to engine thrust
air-breathing hypersonic flight vehicles (AHFVs) can
0 : trim condition
carry more payload than equivalent rocket-powered
∞ : free stream condition

Since testing via ground facilities gives limited infor-
1 INTRODUCTION mation about the behavior of the air-breathing scram-
jet technology, flight tests at hypersonic conditions are
A recent multi-year research program of NASA, needed for a broader understanding. For this pur-
Hyper-X, focuses on one of the greatest aeronauti- pose, several flights at Mach 7 and Mach 10 have been
cal research challenges, air-breathing hypersonic flight planned. For each flight, the profile (see Figure 2) has
(AHF).1–5 The main task is to design and build ve- a ground track of 400 miles. X-43A rides on the first
hicles that will fly at speeds of Mach 7 and 10 with- stage of a Pegasus booster rocket, which is launched by
out using any rocket noting that the world’s fastest a B-52 plane. The booster accelerates X-43A to the
air-breathing aircraft for the time being, the SR-71, test conditions at about 100,000 feet and then sepa-
cruises slightly above Mach 3 and the highest speed rates from X-43A. After this point X-43A flies under
attained by NASA’s rocket-powered X-15 was Mach its own power and glides back to earth.
6.7, back in 1967. The prototype (X-43A) developed
for this task is depicted in Figure 1. Based on this pro-
totype, the first goal of the Hyper-X Program, which is
conducted by the Dryden and Langley Research Cen-
ters of NASA, is to demonstrate and flight validate key
propulsion and related technologies for air-breathing
hypersonic aircraft.

Figure 2: Hyper-X flight trajectory (NASA LRC

Once separated from the booster, the control system
of X-43A is required to regulate a strict flight con-
dition for the duration of the test flight and to per-
form a series of maneuvers to identify hypersonic flight
characteristics. The Dynamics and Control Branch
of NASA Langley Research Center (LRC) is respon-
sible for development of the X-43A control system.
The non-standard dynamic characteristics of X-43A
Figure 1: X-43A (NASA LRC archive). together with the aerodynamic effects of hypersonic
flight make the control system design very challeng-
The interest in AHF is based on the advantages it of- ing. Moreover, the wide speed range of operation and
fers, namely increase in payload capacity and propul- the lack of a broad flight dynamics database suggest
sion efficiency, expansion of launch windows, and re- that plant parameter variations and uncertainties are
duction in construction and operation costs.1, 2, 4 The the main issues to deal with in the controller design.
key technology behind AHF and its benefits is usage of
the so-called scramjet (supersonic-combustion ramjet) Although there have been other research programs re-

American Institute of Aeronautics and Astronautics

the IFTV program was cancelled in August 1967 due to technical difficulties such as 2 A BRIEF HISTORY considerable problems and delays encountered in the ground test program due to inlet–combustor interac- tions.6. As can be oratories (GASL) under supervision of Ferri9. The key ideas that enable issues. In Section 5. which can be found in Section 7.. since the evolve- ment of the hydrocarbon-fueled conventional ramjet Another major test program was the NASA Hyper- (CRJ) engine concept. Al- though a non-powered flight test vehicle was launched in January 1967. The paper is concluded with a discussion Force funded Scramjet Incremental Flight Test Vehicle of the future trends in the air-breathing hypersonic (IFTV) program. another model known as the aerothermody- results due to the lack of demonstrated scramjet com. namic integration model (AIM) was ground tested and ponent performance and the lack of test data on hy. programs.72 M in 1967.9.g. The PIBAL and ters and the corresponding control laws are presented GASL scramjet engine programs inspired the U. speed range scramjets was approached by other re- acteristics of air-breathing hypersonic aircraft in Sec. the later studies ter. as a flight-weight hydrogen-cooled structure. 19 fo- understood from the above. there are several aspects cused on the development of fixed geometry scramjet of the research on hypersonic flight. Air in Section 6. we describe the basic dynamic char. tention on hypersonic cruise missions.g. modeling. traction ratios. hydrogen-fueled scramjet engines led to increasing at- proaches to the same problem adopted by NASA cen. Hypersonic air-breathing propulsion has been studied by NASA for more than sixty years. the emphasis of NASA research shifted to Meanwhile. (see. 10. were outlined in early 1960s. 9 Reviews of the early works sonic Research Engine (HRE) program which began on CRJ development and the early design approaches in 1964 aiming to flight test a complete. wall cooling and frictional losses. 3 American Institute of Aeronautics and Astronautics . Soviet Union. and France. In this work. several times with a dummy test ramjet attached to it It had been proposed to add heat directly to a super. we review approaches proposed for The promising high-speed performance of the the AHF control design problem in academia. mainly in the United States. A structural full-scale HRE model known as the mance of the scramjet engine would exceed that of the structures assembly model (SAM) was tested at NASA CRJ somewhere in the speed range of Mach 6–8 and LRC and the original model was enhanced by using would be superior at higher speeds. the feasibility of develop. e. vehicle acceleration from boosted speed of 5400 ft/s to Some final remarks are made in Section 8. sonic stream. The ap. The AIM ground tests conducted from September on high-speed engine development were concentrated 1972 to April 1974 produced a comprehensive database on the scramjet engine. In this paper. The design of high performance wide gies in Section 2. Glenn Research Cen- personic CRJ engines. flight weight scramjet research engine on the in. such as the technical hurdles about fuel injection and mixing. 14 In the late 1940s. wide speed ranges of operation in these programs are thermal compression and variable aerodynamic con- After presenting a brief history of the AHF technolo. X-15 rocket-powered research plane. which was initiated in April 1965. by means of a standing wave. gives an adequate perspective of the current Brooklyn (PIBAL) and General Applied Science Lab- state of the hypersonic flight technology. at least 6000 ft/s using four hydrogen-fueled scramjet modules located around the central vehicle body. using different techniques ical models in the literature for the flight dynamics of such as using dual-mode engines and utilizing mechan- AHFVs. 13. while. we make a distinction between ically variable geometry engines as well.. 18.7–12 ). 24–26 X-15 flew ing a scramjet engine was attracting attention as well.1. described briefly dynamics Laboratory of the Polytechnic Institute of above. 16. and control range. After the cancellation of the IFTV and HRE flight test and nozzle performance.9. Section 4 is devoted to the existing mathemat.lated to or independent of NASA.9 transportation and possible control design directions The purpose of the program was to demonstrate the to meet these trends. the The scramjet engine programs conducted in Aero- current Hyper-X Program of NASA. in scramjet.S. regeneratively to achieving increased flight speeds can be found cooled. analyzed in the NASA John H. the former tion 3. Subsequently. performances of CRJ and scramjet engines developing and ground testing new engine models for a were compared. Mach 3–12. 20–23 approaches by academicians and those used in indus- try. this program was terminated without reach- 1946. and reached a record-making speed of 6. 15 The generic problems needed to be addressed ing its main goal due to financial problems in 1968. and it was concluded that the perfor.9.6. Simulta- however these studies could not produce substantive neously. as early as However. we engines that have good performance over a wide speed concentrate on flight dynamics. e. 17 The studies new techniques and developing new structures such on comparison of CRJ and scramjet engines continued. search groups.

In 1996. based on the X-30 configuration. the National Aerospace Plane (NASP) which is “smart scaled” from a 200 foot operational program. During the program.33 Some of the later developments and the corresponding research at NASA LRC are pre. X-43A is a 12 foot-long hydrogen-powered experimen- ate a major hypersonic flight research program includ. which is currently underway. 36 The airframe-integrated engine con. some guidance the SAM and AIM programs has been popular. eighteen years after the cancella.9 tal engines were tested at NASA LRC in the Mach Moreover the AIM program demonstrated high inter. March 1995. Air Force and NASA to initi. 37–40 The focus of the and un-powered flight tests at Mach 7 and Mach 10.35. X-43B is a 35’–45’ reusable demonstrator vehicle with stage-to-orbit (SSTO) experimental aircraft. and has become a preferred configuration in The exclusive Hyper-X program. other systems. The newly developed scramjet engine technologies of X-43C. Shortly. The vehicle similar to X-30. X-43D is a concept to be used for flight testing hydrogen fueled 4 American Institute of Aeronautics and Astronautics . A engine which will be used to accelerate the vehicle from large number of newly developed modular experimen. NASP program was to build an air-breathing single. and structural dynam- ics. (HySTP) in 1993 and then completely terminated in ing Laboratory (HFL) in November 1991. and certain perfor- has been utilized by the French ESOPE engine and mance and stability issues were analyzed.9.44–51 In spite several Russian engines9. hicle system development rather than an incremental ration.27 between the airframe. 4–7 regime.7–7). the other phases of HyFLITE. is currently foreseen to be completed by 2017. and incremental hypersonic technology program. specifically the development of cles are planned to be constructed and flight tested. 28–30 as well. A review of the HRE program is brought into picture with emphasis on the interactions presented in. 41 The NASP program led to ex- nal thrust performance. which would be used for hypersonic flight test. 38 In the first two of the three not successful due to HFL system problems. and thermal the vertical forebody compression and uses in-stream management are needed before building an operational struts as housings for distributed fuel injectors. which were all cancelled.4 the airframe-integrated engine concept. the NASP program was first reduced in performed four major flight tests on axisymmetric scope to the Hypersonic Systems Technology Program hydrogen-fueled engines utilizing a Hypersonic Fly. In. and and control schemes were designed. which is considered recent years. four small scale (X-43A. engine. the X. in 1986. tal vehicle with a wing span of 5 feet (see Figure 1) ing flight tests. 43 the dynamic characteris- actions between fuel injector stages. tensive studies on dynamic modeling and control of sonic to subsonic mode of combustion. smooth transition from super. 9.S. this concept utilizes the inlet sidewalls program that further development in scramjet propul- to produce extra horizontal compression in addition to sion systems. and combustor tics of AHFVs with airframe-integrated engines were design approaches. as the last stage preceding prototype development for AHF. sion modes from subsonic flight through full scram- ing and demonstration. concept. strong inter. materials. 33–36 Superior to the axisymmetric HRE technology program. The aim of HySTP was building and third of these tests were supported by the French up and testing a Mach 10-15 scramjet propulsion in a aerospace institute ONERA. and the need for flight testing and flight demonstrating these tech- nologies led the U. jet operation (Mach 0. Mach 5 to Mach 7 during the flight test. 37. 9. X-43C is a 16 foot-long there had been extensive development in the rect.6. X-43D) and one full scale demonstrator vehi- 1970s and early 1980s. The second to funding problems. and the fourth was a test project called the Hypersonic Flight Test Exper- joint project with NASA. November January 1995 without conducting any flight test due 1992. a combined cycle engine to be tested in all the propul- 30. hydrocarbon-powered vehicle utilizing a three-module angular airframe-integrated scramjet technology. early works on the airframe-integrated scramjet con- cept can be seen in. 9 Japan. Integration of the scramjet and a test AHFV would be performed at the last stage. X-43B.6. a few The axisymmetric configuration of the HRE used in mathematical models were developed. a rocket three tests were completed successfully producing data powered vehicle would be used to bring the scramjet for both the subsonic and the supersonic combustion system to hypersonic conditions similar to the HRE modes. It is planned to be used in scramjet powered tion of the IFTV program.on inlet and combustor performance at Mach 5–7. the cept has also been studied extensively in Russia and Hyper-X program.9. The original NASP program on the other hand was Another concept NASA started to focus on in 1970s an attempt for a full-scale operational prototype ve- is the rectangular airframe-integrated engine configu. 25.1–5. Later the Rus. 37.42. NASA started a similar but more extensive sented in. It was realized with the NASP configuration. During the NASP program. of the enormous amount of effort and achievement in sian Central Institute of Aviation Motors (CIAM) many aspects.30–32 program with X-15. and February 1998. Although the third test was iment (HyFLITE). the AHFVs as well.

47 For these configurations. The current flight vehicle projects. aerodynamic database devel.scramjet engines at velocities of Mach 15 or greater. gram are presented in. respectively. on the other hand. especially the NASP U. the big picture in as many aspects as possible in a The ground tests. PROMETHEE program focuses on usage of hydro- 43C are currently funded.S. In the rest of namics (CFD) analysis.9. In the with airframe-integrated scramjet engines similar to flight tests. characteristics of these vehicles carefully since they differ in major ways from those of a typical aircraft. with the major hypersonic flight technologies in an attempt to show ground tests performed and the flight tests underway.69 The PREPHA program. Above. it is accelerated to Mach 16 using a large X-30 or X-43A. it is expected to be incorporated with the hy. Center. only X-43A and X. CFD was exten- operational vehicle and whose size is large enough to sively used in design and analysis of scramjet en- operate over all air-breathing propulsion speeds. and opment. computational fluid dy. Amongst them is flight controller design. flight tests is the Russian IGLA program. Kakuda Research is preferred. Successful attempts with IGLA are re- that there is no major difference between the dynamics ported at low hypersonic velocities. we presented a brief history of the air breathing on the other hand. a scaling that would keep the and the extensive series of engine ground tests per- propulsion effectiveness the same and make the flight formed using the very capable facilities in the Japanese tests results for X-43A valid for the operational vehicle National Aerospace Laboratory.4. More detail can be found in the works airframe integration issues.66. The first X-43A flight test on June 2. addressed. 5 American Institute of Aeronautics and Astronautics .000 feet) on the first stage of AIR-BREATHING HYPERSONIC a Pegasus booster rocket. with than hypersonic transportation. 63–65 IGLA We consider generic hypersonic aircraft configurations is a winged scramjet powered gliding vehicle. supersonic. The X-43A project.5 seconds into the mission. After accelerating to the test conditions (Mach 7 or 3 FLIGHT DYNAMICS OF Mach 10 at about 100. and hypersonic regimes. the X-43B AHFVs. In order to design reliable and effective controllers for personic transportation and. cited above and the references therein. in subsonic. How- ever. however.S Air Force currently conducts another scramjet and the Hyper-X programs. ilar to PREPHA that aims to develop enhanced wide range scramjets for AHFVs.68 engine. The X-43C project started only at low level in 2001 and will continue through 2007.64 The other major of X-30 and X-43A since the X-43A model is obtained recent/current hypersonic flight technology programs by photographically scaling each part of an operational are the French PREPHA and PROMETHEE pro- configuration which is very similar to X-30 except the grams. The second flight test provide hypersonic flight speeds or wide speed ranges is planned to be performed in 2003. have all shown that in or- development program. the only current exclusive hyper- teristics of air-breathing hypersonic aircraft with in- sonic flight technology program including ground and tegrated scramjet engines addressed in the literature. in particular.40. propulsion system limited space. we focus on flight dynamics. control. we review the basic dynamic charac- Outside the U. The JAPHAR pro- the first flight tests to be conducted in 2010 and 2016. modeling.. ented. 62 ficient there are other key technologies that need to be Although the HyTech program is currently missile ori. which ended in Finally a large-scale reusable demonstrator (LSRD) 1999 focused on development and testing of hydrogen- vehicle. is an ongoing program sim- X-43D project is in a formulation stage. whose architecture is the same as that of the fueled scramjets.52–61 The X-43A flight test profiles are shown in Figure 2. and simulation studies for the Hyper-X pro. X-43B and LSRD projects carbon fueled scramjets in air-launched missiles rather are scheduled around 2006 and 2010. the booster rocket and then is let to glide down to scram- primary lift generating surface is the body itself due jet operation speeds at which the engine tests are to inefficiency of using a thin wing. the Air Force Hydrocarbon der to make hypersonic transportation feasible and ef- Scramjet Engine Technology (HyTech) program. engine designs. 67 the French–German JAPHAR project. is highly active. X-43A vehicles are planned VEHICLES to perform powered and un-powered flight tests using its scramjet engine and gaseous hydrogen fuel for 5–10 seconds. it is essential to consider the unique dynamic and X-43C projects of the Hyper-X program. is gines and several prototypes were developed includ- planned to be built and flight tested. In this section. It is worth noting performed. For the engine. Among these ing a Mach 3–12 wide range ramjet. 9. the studies that pursued. In this program. the hypersonic research was almost was unsuccessfully terminated after a booster failure solely focused on development of engines that would 13. 43. this paper. 2001 Before the 1990s. gram.

Temperature of the com. shock waves. 75 The most essential parts of the ing moment. the lift while turning the power on causes negative tant materials and active structural cooling must be lift effect together with a large increase in the pitch- utilized. Several flow energizing mechanisms have been the drag. hypersonic speeds cause the so-called sues characterizing the AHFV flight dynamics in more “path-attitude decoupling” phenomenon which can be detail. while passing from the subsonic to supersonic regime dynamic boundary layer thicker. With the maneuverability at lower speeds and keeping the low viscous effects. on the other hand. On ground test facilities. and drag differs significantly from the results of the inviscid analysis The dynamic characteristics of the AHFVs vary much for the aerofoil. which are estimated to that turning the power off causes a sharp increase in be in the range of 2000◦ F–4000◦ F.1 Effects of the Hypersonic Speed flight path. the scramjet en. Viscous effects. These effects have been concluded to AHFV to be considered for the high temperature is. 79 it is pitching moment coefficient. 43. Below. High temperatures on the control surfaces. frame. 73. Moreover.43 This may cause violation of temperature ing the ground tests at NASA LRC that examined the and structural constraints. the effective aerodynamic surface is not speed static stability margins in the conventional range the aerofoil itself but the boundary layer together with and accepting unstable configurations at hypersonic the aerofoil. These effects be- come significant at Mach numbers M > 6 due to the increase in the temperatures behind the normal shock 3. Two of the main The effect of hypersonic speed on flight stability of ve- impacts of the “real gas” effects are the increase in hicles similar to X-30 and X-43A configurations was loading on certain aircraft surfaces and the increase in extensively studied as early as 1970s. pressure distribution.76–78 Hence the actual flight path signifi- “Real gas” and viscous effects become important at hy. temperature resis. basically described as the resistance of the high mo- mentum of the AHFV to the changes in the desired 3. which will be explained in detail in the many aerodynamic and propulsion characteristics still following section. 70 The dy.43. large control deflections. and hence increases as well. variations on a specific path is faster than the lower ve- namic characteristics of the hypersonic vehicle vary locity regimes. be due to the diverging angle between the nozzle and sues are the combustor wall. In. cantly lags the changes in the pitch attitude at hyper- personic speeds. applied by the controller to these surfaces. conditions.2 Variations due to Wide Speed Ranges waves with increasing Mach number. The difference is not only due to the more over the flight envelope than other aircraft due thickening but because of the skin frictions caused by to extremely wide range of operating conditions and non-smoothness of the boundary layer as well. In order to counteract the subsonic AHFV behavior close to the ground plane high operating temperatures. developed to overcome the latter effect. the strong interactions between the elastic air. we analyze the major is. put constraints on the deflections speed. “real gas” effects denote sonic speeds. 70. also affects the landing/take off performance and in- sure. which causes an increase in aerodynamic heating creases the transonic drag. the guidance scheme and the flight remain uncertain and are hard to predict due to lack control system need to be highly integrated in order of sufficiently many flight tests and inadequacy of the to provide a robust stable high performance flight. 72. the speed of the flight condition namics of AHFVs highly challenging.43. and decreased boundary layer at hypersonic flow regimes. The hypersonic configuration needed for efficient AHF gine needs to be operated at a high dynamic pres. Hence the viscous behavior in terms of speeds. Considering this fact in addition to the more significantly over the flight envelope than other variations due to wide speed ranges of a complete flight aircraft due to its extreme range of operating condi. 43. the ground plane. the deviations of the aerothermodynamic properties of the air from the ideal gas behavior.80 The external nozzle configuration and the control surfaces. In order to reach hypersonic speeds. for good hypersonic performance causes flow separa- bustor wall is directly related to the engine operation. tion at lower speeds and a steep increase in the drag Increase in the inlet ramp temperature makes the aero. the other hand. are basically due Hence there exists a trade off between having rea- to the thickening of the boundary layer around the sonable static margins by allowing high static stabil- aerofoil by the large temperature gradients across this ity margins. mass distributions. and the structural dy- namics make the explicit characterization of flight dy. the fore-body inlet ramp.Beside the aerodynamic effects of the hypersonic on the other hand. the propulsion system.43.40. and the sensitivity of the AHF dynamics to the flight tions and rapid change of mass distribution.40. 70–72 Here. It has been recorded dur- and drag. 74 concluded that the static stability margins of such AHFVs generally decrease as Mach number increases. At hypersonic speeds. 80 6 American Institute of Aeronautics and Astronautics .43.

hicle fore-body. Because of the engine-airframe integration issues explained in Section The scramjet engines operate by supersonic combus- 3. the airframe-integrated grated exhaust nozzle configuration generates propul- scramjet engine concept has been studied extensively sive lift and increases the lift-to-drag ratio. in order to produce suffi- non-uniform. 43. the airframe inte- liably.6.. Aside from the airframe-integration issue. and Japan. tutes the back part of the grey section together with the lower boundary of the after-body. the engine inlet needs to capture as much of the airflow under the AHFV surface as possible. there is significant interaction between the propul- tion of fuel in an air stream compressed by the air- sion system and the vehicle aerodynamics. Using hydrogen as the basic Considering the longitudinal dynamics. 43 It is worthwhile to with airframe-integrated scramjet engines similar to note here that the net thrust in AHF is usually a small X-30 or X-43A. The sensitiv- gine.3. propulsion to subsonic speeds reduces the performance signifi. the airframe. and structural dynamics are highly interac- cantly. hence any additional drag sic features of the scramjet engines in order to model source can affect flight performance significantly. it is essential to understand the ba. In the minimum-area cal analysis in. cient thrust for hypersonic flight. The grey ity of the performance to the AOA and the dynamic section under the AHFV body in the front and side pressure is small at high Mach numbers and increases views contains the internal modules.6. strated via the results of a frequency domain numeri- ment with the fore-body flow.3.6. 34. fraction of the gross thrust. As stated in Section 2. of this two-fold contribution to the flight performance. addition to increasing the net thrust. tive. These flows may and react in supersonic combustion. cates a similar effect of the fuel flow rate and diffuser straint on the mixing time due to supersonic combus. and has become a pre. ferred configuration in recent years. the combustor also affect the lateral dynamics if they are laterally needs to be longer. the integration of the exhaust noz.40. and the propulsion system is demon- few degrees from the adjacent modules for better align. Bending of zle with the airframe provides larger net thrust. Russia. Injector distribution relaxes the con. i.g. 73 is based on the potential performance at hypersonic speeds (M > 6). 43 The compression of the flow through the inlet is determined inlet sidewalls produce extra horizontal compression in by the properties of the bow shock wave under the ve- addition to the vertical fore-body compression. tion and shortens the required combustor length. the internal modules and namic pressure also effect the combustion kinetics and the exhaust nozzle are the main components of the en- the exhaust flow/free stream shear layer. Each module diverges a ics. The capture and contiguous with the vehicle undersurface. the preference of scramjet engines to ramjets and 3. the air-breathing scramjet engines pressure acting on the fore-body generates a nose-up burn oxygen scooped from the atmosphere. 80 craft’s forward speed.3 Scramjet Engine Dynamics increase in the gross thrust without significantly in- Since our focus is on hypersonic aircraft configurations creasing the drag that much. FVs is the structural aeroelastic modes. 42. 9. Figure 1 illustrates these components.. e.43. the fore-body and after-body together with propaga- tions throughout the entire airframe affect the flows 7 American Institute of Aeronautics and Astronautics . 34. At low speeds. The preference use of this configuration is essential in AHF.43 This analysis shows that the thrust section of each of the internal modules there exist a and the pressure at the engine inlet are significantly number of wedge shaped in-stream struts that are used effected by pitching control surface.e. the elastic airframe. 81 The AOA and the dy- Beside the fore-body inlet. elevon deflec- as final compressors as well as housings for distributed tions over a wide frequency range.4 Airframe/Propulsion/Strucural Dynamics the other conventional engines at hypersonic speeds is Interactions due to the fact that slowing the high-velocity airstream In X-30 and X-43A. the AOA and this section and the lower boundary of the fore-body the dynamic pressure need to be greater than certain constitute the inlet. the inlet flow fuel for combustion. Since the pitching moment while the external nozzle flow gener- hydrogen and the scooped oxygen have less time to mix ates a nose-down pitching moment. the aerodynamics of the AHFV effects the the engine with the airframe so that the inlet area is propulsion system in several ways. The internal modules are located symmetrically with respect to the The dynamic coupling among the AHFV aerodynam- AFHV plane of symmetry. Because in the U. 43. hence the control tasks for flight and propulsion systems cannot be separated clearly. Another major source of dynamic coupling in the AH- Similar to the inlet. while the exhaust nozzle consti- values in order to provide acceptable performance. The analysis indi- fuel injectors. In and control flight dynamics of these configurations re. system..S. which are determined by the angle of attack (AOA) and the dynamic pressure as well as the free stream characteristics. The front part of as the speed decreases. area ratio changes on the AHFV pitch rate. This is provided by integration of Similarly. Moreover.

51 Even if dynamic interactions Therm. the state. conventional control methodologies that tural vibration frequencies which appear as a result depend on relatively accurate models and a reliable of the requirement for very low structural weight mass aerodynamic database may not be applied and the cor- fraction. and the structural dynamics are aerodynamics.43. In this section we review the AHF literature benefit of a vast statistical database of actual flight where the distinct characteristics of hypersonic flight test data.4 are some of these unpredictable x is a vector of dimension 5–9 which contains informa- behaviors. Further information on The dynamic characteristics of AHF described in Sec- aeroelastic effects and mode shapes of AHFVs can be tion 3 make the modeling and control of AHFVs very found in. pitch rate. y ¯ = h(x. As a result. especially for the flight using intelligent. must proceed without the problems. Meas. the uniform pressure distribution. 89 Furthermore Structural Dynamics Bending Inlet/Nozzle Conditions Engine Dynamics Thrust/ Lift AHF Dynamics Attitude&Loc. have assumed parameter variations conventional aircraft and the design of their control and uncertainties to be small. the unique Figure 3: AHFV dynamics. where x. and the propulsion sys- illustrated in Figure 3. The nominal longitudinal models. the elastic mode variations. the engine. respectively. namic characteristics are very hard to predict while dynamic performance. ited. in general. non. Due to the strong coupling between the namics. the modeling and control techniques used for conventional aircraft are inadequate if feasible at all. u. x˙ = ¯f (x. pitch attitude. and propulsion uncertainties that are a result of insuf- ficient flight test data and inadequacy of the ground The majority of the AHF dynamic models in the lit- test facilities. Section 3. 83–88 The interactions among the flight dy- challenging. Hence. 82 Accurate determination of the structural responding uncertainties need to be dealt with during elastic modes is critical for flight control. Engine Flight Mathematical models incorporating the interactions Control Control within the AHFV as well as an integrated airframe- AHFV engine control system are needed. u) (4. are described Unpredictable aerodynamic and thermodynamic be. wind tunnel data. however.S. However the non-uniform aerodynamic heating in AHF. works in the AHF control literature have ignored the ful CFD codes available today.. Thermodynamic Effects are ignored and parameter variations and modeling uncertainties are assumed to be negligible — an in- valid assumption as demonstrated in51 —.1) other uncertainty source. input.43. unconventional compos- ite materials used in building the airframe. Typically. the development of coupling effects. cant nonlinearities had to be introduced making them and flight test data. distinctly different from the conventional flight control personic aircraft. erature consider only the longitudinal motion. precise control of the AOA. 51. signifi- such as U. 82 The propulsion sys. Elastic-rigid body interactions the ability to experimentally determine them is lim- are also significant in AHFVs due to the low struc. Aerodynamic the large variations and uncertainties associated with Forces&Torques Flexible Modes the dynamic characteristics over the flight envelope and the corresponding robustness issues need to be taken into account.2 can also be considered as uncertainties in and altitude. there are many aerodynamic have been incorporated in the modeling process. The mass property variations discussed in tion about velocity. aerothermodynamic characteristics of hypersonic flight explained in Section 3. in state space by equations of the form haviors due to hypersonic speed and airframe/ propul- sion/structural dynamics interactions constitute an. Design and development of hy. or robust control. The input u is 2–3 dimensional vector modeling. adaptive.1 need to be considered. FLIGHT DYNAMICS tions and uncertainties in the shapes and natural fre- quencies of the elastic modes.43. the airframe. 47. AOA. and y denote.47. Al- though a number of studies on the control of AHFVs 3. and the 4 MODELING THE HYPERSONIC shell type structure of AHFVs cause significant varia. and uncertainties ex. and the output of the AHF system. most of the Despite the unprecedented capabilities of the power.5 Databases and Uncertainties do consider all of the challenges above. many of the critical aerody.43. plained in Section 3. Air Force DATCOM. whose entries are adjustable control coefficients. u) tem perturbations that indirectly impact the longi- tudinal dynamics.through the inlet and the exhaust and hence the aero. tem. In general. such 8 American Institute of Aeronautics and Astronautics . and have considered systems depends on the use of empirical estimates of only some of the aerothermodynamic characteristics aerodynamic and engine data obtained from sources due to hypersonic speed. Even in these cases.

Numerical values of A and ory. be ing the dynamic model of 90. These effects and vari- x = [ V (ft/s) α(deg) q(deg/s) θ(deg) h(ft) ] . degree of freedom (DOF) nonlinear rigid-body simula- is described by tion of the conical accelerator vehicle using a software program to optimize simulated trajectories(POST). further anal- ceptably estimated—. 91 is significantly different T from the rectangular airframe-integrated engine con.82 the impact of the propulsive perturbations to the system and/or fuel and air flow rates in the on the pitching moment is focused among the cou- integrated engine. More specifically. and the engine modules are placed cylindrically all around the and v a random vector of dimension 28 following a nor- body.3) where f and h are smooth nonlinear (vector) func- tions. 1}. m re2 lated as uncertainties and incorporated into the model L + T sin α (µg − V 2 re ) cos γ (4. 94 the system variables are defined to The winged-cone configuration considered in deriv. 99 without elaborating their behavior. v) (4. in its most general form. carries information about the altitude and the velocity ∆B = 05×2 and ν is an uncertain parameter. recent and ongoing developments in non- x˙ = Ax + Bu (4. A drawback of the linear approaches above is that the erator configuration. are represented by a set of random T u = [ δe (deg) ηf ] variables in the context of stochastic robustness95–97 as an alternative to the classical (linear) robustness the- is developed and studied. They differ in the coupling effects realistically is limited. ations. in.5) In. however. In. A widely used longitudinal model in the literature is the rigid-body model of 90.93. u(t). It typically where ∆A = ν [ 0 0 1 0 0 ] [ 0 1 0 0 0 ].2) as γ˙ = − (4. For this of the AHFV. In this B for certain flight conditions are derived from a 6- context. Later can be linear or nonlinear. The coupling effects for the linear T cos α − D µg sin γ V˙ = − (4. figuration in that the forebody is axisymmetric and u = [ δt T δe (rad) ] conical.4) by defining a cover function carrying informa- tion about the flexible modes. A nonlinear where A and B are system matrices that depend longitudinal model for a specific flight condition based on the flight condition —the Mach number and the on90.2) linear control would relieve this burden. 94 In this model the coupling altitude— and effects and other possible variations from the nominal T system are also considered. 91 for the winged-cone accel. uncertainty in Cmα .82 the coupling between elastic and rigid structure extended to include uncertainties and external distur.7) mV V re2 x˙ = (A + ∆A )x + (B + ∆B )u (4. The longitudinal dynamics are explicitly given by namics. modes is included in the frequency domain equivalent bances. and propulsion properties assumed. the system model. measurable entries of x.8) 9 American Institute of Aeronautics and Astronautics . One way to the method used for modeling and linearization—if the eliminate this drawback is introducing nonlinear mod- model is a linear one—.82 They are formu. or some ysis has been performed for the special case of (4. of (4.4) is scaled so that ν ∈ {0. In91 Although using a nonlinear model puts limitations on the linear model the designs that can be used to control the system in general.as elevon angle and fuel equivalence ratio. and structural dynamics of the ve- hicle are negligible. and v is a random vector with normal proba- where d is a vector of dimension 2 which represents bility density function representing the uncertainties the atmospheric disturbances and Ex is a 5 × 2 system in the parameters of the dynamics.98. propulsion. The output y can be the state x pling effects and this effect is addressed as parametric itself —if the state x is fully measurable or can be ac. where ∆A and ∆B are additive uncertainties respec- mine the aerodynamic and propulsive power supplied tively.91 the atmospheric affects such as turbulence are included and the model (4. In this direction. Several different formulations of capability of the model to represent the dynamics and this model are available in the literature. (4. that deter.93. Moreover the model is derived assuming that mal distribution with mean 1 and standard deviation the body is rigid the coupling effects among aerody. the nozzle section is a cone frustum. 91 is presented in. v) x˙ = Ax + Bu + Ex d (4.92 ˙ x(t) = f (x(t). x = [ V (ft/s) γ(rad) h(ft) α(rad) q(rad/s) ] .4) T measurements giving information about x.6) model (4. The nominal system functions ¯f and h ¯ case study. mass.2) are later considered in. els that carry more information instead of linear ones. be constant throughout an individual process. v is assumed to matrix depending on the flight conditions. environment.2) is generalized to y(t) = h(x(t). and the aerodynamics. 0.4) h˙ = V sin γ (4.1. and their domain can be in.

. . be rewritten as and simple lumped-mass model characterization of the vibration modes in the airframe structural dynamics.12) ∆η˙ 0 Z t     ∆M∞ hI (t) = (h(τ ) − hd )dτ (4. the thrust T . 47 The focus of this model is dy- so-called triangular form.5) is too general to form a hypersonic-vehicle model derived for the X-30 con- basis for control design. mulated as functions of the system parameters they depend on as well as some of the entries of the ran. Although the models above are simple and suitable for dom uncertainty vector v. they do not de- above symbols. x3 . a∞ . . x2 . clusive and more realistic mathematical model in the literature is the analytical aeropropulsive/aeroelastic The nonlinear model (4. which is suitable for a large namic couplings and control system integration. signs based on these models in Section 5. CL . x2 . v) moment M .5) and (4. .5) can aero/thermo flow analysis of the propulsion system. Cm (α). v) The elastic motion is characterized by a single body x˙ 2 = f2 (x1 .10) are expressed in the ∆QT ∆η˙ 10 American Institute of Aeronautics and Astronautics . . see Nomenclature at the beginning of scribe the AHF dynamics detailed enough. The set of control design procedures in the literature as corresponding approach involves a two-dimensional well as being general enough to describe many physical Newtonian aerodynamic analysis. 47 are summarized as follows: where fi . v) + g(x1 . . Iyy . x2 .     ∆M∞ ∆XA instead of assuming perfect desired setting tracking  ∆α   ∆ZA    and defining   = A  ∆q  + Ac ∆δe ∆MA   Z t ∆η ∆QA VI (t) = (V (τ ) − Vd )dτ (4. A more practical model is the figuration in.. In. x3 . (4. Cm (δe ). x3 . and the radius re from the Earth’s center x˙ 3 = f3 (x1 . a one-dimensional systems.93. more clear with presentation of the robust control de- CD . x2 . xk . x˙ 1 = f1 (x1 . v) + g(x1 . The use of the models (4. A more ex- the paper. The control effectors considered are the pitch-control surfaces for the aerodynamic system . x2 . 94 the model parameters m. v) where the lift L. xk . In this form. v)u T porates most of the issues covered in Section 3 and x = [ xT1 xT2 ··· xTk ] summarized in Figure 3. the drag D.46.14) re = h + Re as well as the stochastic robustness context will be In. . x2 . x2 .10) x˙ 1 = f1 (x1 . The derived force and mo- y = h(x. the ∆MT   ∆T0 ∆η longitudinal dynamics (4. . and h are smooth nonlinear vector and X = XA + XT + XE matrix functions. the system equations (4.11) and the fuel flow and diffuser area ratio adjusters for the engine. The mathematical model derived incor- x˙ k = fk (x1 . S.13) ∆XT · ¸  ∆α  0  ∆ZT    ∆A¯D   = T  ∆q  + Tc where the subscript d stands for the desired value. For the definition of the developing control design techniques.6)–(4. v) bending mode. x3 .94 considering the Z = ZA + ZT + ZE engine throttle dynamics to be of the form M = MA + MT + ME δ¨t = δtcom Q = QA + QT + QE where δtcom is the throttle setting command signal.14) are modeled as where 1 2 L = ρV SCL x1 = [ VI V γ] T 2 1 2 x2 = [ δt hI h T α] D = ρV SCD 2 = [ δ˙t T x3 q] 1 2 T = ρV SCT u = [ δtcom δe ] T 2 1 2 M = ρV S¯c (Cm (α) + Cm (δe ) + Cm (q)) and the random uncertainty vector v is defined the 2 same as before.9) triangular form as q˙ = M/Iyy (4. x2 . v)u(4. α˙ = q − γ˙ (4. respectively. CT . v) ment equations in46. ρ. and Cm (q) are explicitly for. the pitching x˙ 2 = f2 (x1 . . g. c¯.

89 The equations derived in89 include the pressions for all these system matrices.       ∆M∞ 1 0 0 0 0 0 0 ∆XE · ¸  ∆α   0 1/a∞ 0 0 0 0 0   ∆ZE    ∆A¯D     = E  ∆q  + Ec 0 0 1 0 0 0 0  ∆ME   ∆T0 T3 =   ∆η 0 0 0 0 1 0 0  ∆QE   ∆η˙ 0 0 0 0 0 1/∆τ1 0 where the notation given in the nomenclature is used. B = B δpi Bi . u =  ∆A¯D  variable control designs.46. structural vibration mode shapes.e. gether with the couplings among them. T ..46. 47 can be extended to include some other control derivatives. c1 = 2P∞ dh i=1 i=1   0 0 −u0 u0 0 0 0 where  0 0 −g 0 0 0 0     α (rad)  0 0 0 0 1 0 0   h(ft)  m   T1 = 0 0 0 0 1 0 0   qf (rad/s)     u(ft/s)      0 0 0 0 0 0 0     qt (rad/s)     α(rad)  δe (rad)   0 0 0 0 0 0 1   .y =  P2 (atm)  x =  θ(rad)  . which can be used in robust multi-   x =  ∆θ (rad)  . 47. Based on the force and moment equations summa. Although the integrated scramjet engine dynamics and rized above the AHFV longitudinal equations of mo. the stabil. and development of  ∆u (ft/s)    uncertainty models for AHFVs. B = T2 [ Ac Tc + Ec ] i=1 i=1 S = A+T +E np np X X M∞ dP∞ ¯+ C=C δpi Ci . TC and EC are 4 × 2 and contain relatively minor dynamic effects using the equations of engine-thrust and external nozzle control derivatives. noting the similar- 0 0 0 0 0 0 0 ∆τ1 /m ¯ ity between the state and input vectors of (4. y = Cx + Du. the following generic model is considered in:51 1 11  c S  A = T1 + T2  1 21 S  T3 x˙ = Ax + Bu. discussed. elastic deformation.15) and 11 American Institute of Aeronautics and Astronautics . The uncertainty sources have already been ∆τ1 η˙ (rad/s) discussed in Section 3. and E are 4 × 5 system matrices whose ele- ments are the stability derivatives of aerodynamics. All the elements in the second and the model derived for generic elastic hypersonic flight ve- forth rows of T and TC are zero. and 3-DOF point-mass- respectively. engine-thrust. AC The longitudinal dynamic model for the X-30 config- is a 4 × 1 vector whose elements are the aerodynamic uration in46. Detailed analytic ex. rotating machinery. (4. fluid ity and control derivatives can be found in.16) c1 S31 np np c1 S41 X X A=A ¯ + ¯+ δpi Ai . tion is based on a Lagrangian approach and the signif- ditions. i. the hypersonic aerodynamics are modeled to some ex- tion were derived and expressed in the state space form tent. 89 the parameter vari- x˙ = Ax + Bu (4. hicles in. effects of rigid-body motion.15). ¯ + D=D δpi Di . atmospheric con. 0 0 0 0 0 0 1/∆τ1 A. and Mach icance of selected terms in the derived equations are number. and external nozzle. respectively. Elaborating on some of these  c S  sources. The deriva- cle geometry and mass properties. u =  ∆A¯D   0 0 0 0 0 −ω12 −2ζ1 ω1    M   q(rad/s)  m ˙ f (slug/s)        The (lb/ft)  0 0 0 0 η(rad)   nx (g)  1/m ¯ 0 0 0  η(rad/s) ˙   nz (g)  0 1/mu ¯ 0 0 0    A and B are the system matrices carrying the stability T2 =  0 0 0 0    and control derivative information. 47 These flow. They are similar  0 0 1/Iyy 0    to the system matrices of (4. kinematic equations. and checking whether these    ∆q (rad/s)  ∆T0 (◦ R) models satisfy the usual assumptions in robust con-   ∆τ1 η (rad) trol theory. motion. and Earth rotation to- expressions are basically nonlinear functions of vehi.15) ations and the modeling uncertainties are neglected. and the coupling issues in AHF dynamics are as well as analytically elaborated in. These latter issues are considered in detail in model- where ing in.5. specifically for the X-    ∆α (rad)  ∆δe (rad) 30 configuration. wind.51 The focus of 51 is determination of the sources   ∆h (ft) and the extents of uncertainties.

i = 1. desired trajectories during a time interval of interest. Having covered all the main models for AHFVs in the literature above. Especially.40. demonstrated in Section 4. Ci = ∂δpi . Di = ∂δpi . A. δp5 = sues in longitudinal AHF dynamics. cerns in reaching this goal: Stability.¯ D¯ are the effects of AOA and Mach number on the propulsion nominal values of A. . The main models for longitudinal AHF dynamics that ical programs. are considered in the engine model.(4. the mention the Generic Hypersonic Aerodynamic Model existing control design attempts can be thought as Example (GHAME). concluded in51 that some of the general assumptions on uncertainties in robust control theory are violated. and a simplified switching of y) to follow desired time-trajectories by adjusting turbojet-ramjet-scramjet engine model. . In.14). In this section. we see that a comprehensive analyti- p1 = P∞ . (4. certainties can be represented by a set of uncertain The available theoretical studies in the AHF control parameters which does not need to hold at all. which can be assumption of invariance of the number of unstable defined as computation of the command signals to be poles.4). This simple enough to form a basis for control design pur- leads to the corollary that rigid body assumptions ig. Nevertheless. able. and the control problem is expected to become more complex considering the Beside the analytical AHF models above. it is worth to lateral-directional characteristics. 100 an extensive computational the initial steps in solving the integrated AHF control model. ¯ C. The goal of the control design based on anyone of equilibrium model and the second a model obtained these models is to force the system outputs (entries using analytical programs. cally due to the lack of full 6-DOF models which are ditions on their magnitudes are hard to satisfy. and scramjet for V ≥ 6 M.16) is applied to the control effectors of an AHFV in order based on the assumption that the uncertain parame. δp3 = ¯ . is ana- the uncertainties need to be clarified. The other two system matrices. as an example. The tem outputs are required to be close enough to the de- engine mode switches to turbojet for V < 2 M. derivation of the model (4. a set of control coefficients (entries of u). p5 = ∆τ2 cal model expressing all the dynamic characteristics of P∞ − P¯∞ ω1 − ω¯1 ζ − ζ¯ AHF presented in Section 3 in detail is still not avail- δp1 = ¯ .16) ¯ 1 ∆τ ¯ 2 ∆τ covers almost all significant characteristics of longitu- dinal AHF dynamics.90. (4. This is basi- so large over a wide frequency range robustness con. (4. Moreover the modeling and control of longitu- noring the aeroelastic modes will result in greater and dinal AHF dynamics itself is complicated enough as unacceptable uncertainties. to make certain flight condition parameters to track ters are mutually independent and the dynamic un. D. δp2 = . we review the attempts in the litera- merical results do not violate the less conservative ture to solve the AHF control problem. the additive plant uncertainty ¡ ¢ ¡ ¢ 5 HYPERSONIC FLIGHT CONTROL: ∆a = C(sI − A)−1 B + D − C(sI ¯ ¯ −1 B − A) ¯ +D ¯ THEORY AND SIMULATIONS is shown to be unstable which violates the frequently made stability assumptions although the limited nu.51 the case with np = 5. ramjet sired ones while guaranteeing the system parameters 12 American Institute of Aeronautics and Astronautics . Similar to namic models are based on a rigid body configuration any other control problem. P∞ ω ¯1 ζ (4. design methods. the effects of δp1 use of the models above will be more apparent in the and δp2 in the frequency domain are demonstrated via next section where we review the AHF control designs numerical analysis results.¯ B. (4. Specifically.98 Moreover. It is literature focus on the longitudinal dynamics ignor- also remarked in51 that the uncertainties appear to be ing its lateral-directional counterpart. C and D have for 2 M ≤ V < 6 M. and erator. . The validity and lyzed further and. it is based on these models.16) addresses a portion of is- ¯ ∆τ1 − ∆τ1 ¯ ∆τ2 − ∆τ2 δp4 = . (4. although the concerns in51 about where the over-bar indicates nominal values. p4 = ∆τ1 . GHAME is suitable for simulation and numerical ∂A the corresponding fractional variations and Ai = ∂δp i . there are three main con- with a geometry similar to that of the conic accel.5). p2 = ω1 . GHAME is composed of five data sets: Two problem. poses. B. Nonetheless each of the models (4. Based on these results. The the sensor structure information. two aerothermodynamic models first of appear in the literature have been presented in Section which is a simple convection-radiation heat flux and 4. .2). 91 The engine modules are assumed to be robustness. The time-trajectories of the actual sys- wrapped around the lower surface of the fuselage. performance. intelligent controller synthesis rather than analytical ∂B ∂C ∂D Bi = ∂δpi . (4. aerodynamic models one of which is based on several empirical data sources and the other based on analyt. p3 = ζ. np de.15).16). In terms of con- note the uncertain parameters in the system. pi . C. The aerody. δpi are trol.3).

the con- troller for the nominal plant results in significant sen- trol problem is defined as stabilizing the vehicle and sitivity to external disturbances even in the absence tracking the velocity (V ) and altitude (h) commands of the input modeling uncertainty. The limit on the to 17 %. however. an There is always a trade-off between performance and H∞ controller for the uncertain plant—considering the robustness in control design for systems with uncer. DK-iteration in. 20 % input uncertainty in the design procedure—.1) hc Among different approaches in the AHF control lit- erature. the AHFV models with structured and unstructured uncertainties.82 the DK-iteration without any dimension con- rect H∞ loop shaping techniques and µ-synthesis with straint is used first to design a µ-optimal controller.91 the designed controllers are numerically analyzed of robustness with undesirable side effects on perfor. Frequency varying weighting is used for promises a controller with very desirable properties. 99 the DK-iteration using residue and Hankel norm ap- such as H∞ techniques and µ-synthesis. are applied to proximation methods. controller of fixed order. which is and nonlinear model based. a linear simula- tion of the system is used and the entries of the exter- nal disturbance d in (4.91 The design has produced three Using model reduction methods similar to those used in. this trade-off is a µ-optimal controller obtained using DK-iteration. However.. system than performing the DK-iteration and then reducing uncertainty due to parameter variations from vehi. The resultant controllers have the com- mon property of being highly conservative in terms In. satisfies the require- AOA (α) deviation from nominal less than 0. the controller di- cle acceleration are represented as a 2 × 2 diagonal mension is fixed a priori and a fixed-order µ-synthesis 20 % multiplicative input uncertainty. The use of the H∞ con- some small corruption due to sensor noises.5◦ and ments provided that the input uncertainty is reduced minimizing the control power use. In the linear model based obtained by truncating the 18-dimensional result of approaches. For the time domain analysis. We can divide the available robust AHF control ap- respectively. Although it the input u. 700 ft.e. (5. In that paper. AOA minimizes the perturbations in the integrated engine inlet and nozzle conditions discussed in Section 3 which affect the propulsion performance. The robust control problem is approached by both di. y x˙ c = Ac xc + Bc  Vc  . and tainties and disturbances. V and h.99 The detailed scheme of the reformula.3) due to turbulence are im- The robust control design for the rigid conic accelera- plemented as outputs of Dryden turbulence filters fed tor configuration in91 is a milestone among the linear with white noises.to be bounded within allowable margins irrespective of controllers: An H∞ controller for the nominal plant— the modeling uncertainties and external disturbances. The simulation results demonstrate model based approaches. all the robustness and performance requirements while i. while constant weighting is used for other state requirements such as knowledge of an initial stabilizing and input entries.3) with the assumption that y ∼ = x. ignoring the 20 % input uncertainty in the design—. main.4). The tracking procedure is used.91 this controller —of order 47— is truncated to 13 American Institute of Aeronautics and Astronautics . which are required to track certain desired the procedure itself is hard to implement due to some values. The H∞ controller highly accurately while keeping the magnitude of the for the uncertain plant. The nom.98. In this technique. robust control approaches are more realistic u = Cc xc since they take into account modeling uncertainties. The procedure employs a mixed and stability performance requirements are converted µ-synthesis method called DGK-iteration102 in order to weighting function conditions for the state x and to reduce the robustness conservatism. The focus of this work is use of a special in the robust control context in order to apply de. and compared in both frequency domain and time do- mance. the order of the resulting controller. all the states are available for measurement with the H∞ controllers are not. much more significant in AHF control due to the fast Each of the three controllers is of the form time varying nature of the system and the large size   of anticipated uncertainties and disturbances. where Vc and hc are command signals for V and h. rather tion can be found in. technique for designing fixed low order controllers. A similar robust control approach to the same control inal flight condition is assumed to be V = 8 M and problem is conducted in82 based on the uncertainty h = 85. based on the that the µ-optimal controller is successful in satisfying linear model (4.91 In this formulation.101 sign controllers that would provide H∞ -stability and which do not have implementation problems caused by µ-optimality. The control problem is reformulated model (4. classical techniques of robust control. Dimension of xc is 12 for the H∞ con- proaches into two subcategories as linear model based trollers and 13 for the µ-optimal controller. high system dimensionality. In.

5) with dynamics (4. namic signal to ensure stability and desired shape of bustness95–97 has been introduced as an alternative to transition. This as already shown in (4. mate Nv iteration. in terms of robustness in addition to taking into ac- count the system dynamics and the design require- Another H∞ based approach to control of the linear ments in detail. let us denote the plant ∆us = −kR KLQ ∆xc (5. The focus of the approach is de. and the H∞ techniques are used to de. a fixed fifth. vNv }.5) by Gv . 91 is presented in.12). in93 and. The robustness of the ∆xc = [ (∆x)T VI hI ] . All the uncertainty information is collected in a ran. The essential part of control design and analysis in93 is dom vector v and inserted in to the nonlinear model the design and analysis of the component ∆us .5th–9th stabilizing controllers. u = u0 + ∆u. The simulation results x˙ = Fx (v)∆x + Fu (v)∆u (5. which does certainty vector v reliably. In the cases where troller is tested in93 using numerical stability analysis 14 American Institute of Aeronautics and Astronautics .4) in103 demonstrate the effectiveness of the new H∞ based eigenstructure assignment design compared to where Fx (v) = ∂f /∂x. Performance of the full order controller. parameterizing the uncertainties and forming the un- proach is based on the simple form (4. this ap. .103 However. v2 .82 The simulation results demonstrate is used as a practical alternative to PV (K). and u0 is the nominal trim value of the control signal that is assumed to be available. the superiority of the fixed order controller to the trun- cated controllers as well as its successful performance Nonlinear model based (stochastically) robust control and robustness properties which are close to those of techniques have the benefit of being less conservative the full order controller. kR is a scalar constant J(Gv .93 the state equation in velop an alternative to the Shapiro eigenstructure as. is selected from V and the probability esti- troller as the initial stabilizing controller of the DGK. . K) pr(v)dv (5. is acceptable —in terms of stability and performance T requirements— and 1 otherwise. pr(v) R is the probability density function of v and stability metrics defined using the stochastic ro- satisfying v∈V pr(v)dv = 1.3) Nv i=1 troller are illustrated and compared via simulation re- sults as well in. . ple set {v1 .2) the design parameters determining the LQ cost func- v∈V tion to be solved to obtain KLQ based on performance where . ∆u is further composed as It has already been mentioned in Section 4 that it is better to use nonlinear models instead of linear ones ∆u = ∆ud + ∆us in order to represent the dynamics and the coupling effects in detail. all the nominal values of the system parameters of the model modeling uncertainties are assumed to be parametric (5.94 respectively. Later. Since the classical robustness theory where ∆ud is the increment needed to bring the sys- has been developed for linear systems and does not fit tem to a new desired trim condition and ∆us is a dy- well to nonlinear models the context of stochastic ro.2) cannot be computed analytically. (4. In. The acceptability of using a certain controller K to control Gv is quantified via a where KLQ is the LQ optimal gain matrix obtained by binary indicator function J that is defined as follows: minimizing a LQ cost function. In this context. A genetic algo- Z rithm is used to select the robustness coefficient kR and PV (K) = J (Gv . about the trim state x0 as coupling the phugoid and short-period modes of (4. modeling uncertainties are preferred to be expressed in terms of uncertainty transfer functions and matrices. K) used for robustness purposes. sampling.14) uncertainties. where Nv is large enough for nique of 101 together with the truncated fifth order con. and the fixed fifth-order con. These techniques are ap- not take into account the disturbances and modeling plied to the longitudinal AHF models (4. the 1 X truncated controllers.5) and (4. Note that the smaller bustness notions (5. ∆ud is computed algebraically in terms of the classical robustness theory. x = x0 + ∆x. a random sam- order controller is obtained using the special tech. PˆV (K) = J (Gvi . the traditional Shapiro design.5). K) is equal to 0 if the closed loop system (Gv .13).2) using this alternative design. The designed con- PV (K) is the more robust K is. K) (5. controller K over the set V of all possible uncertainty vectors v is measured as and VI .2). (4. Fu (v) = ∂f /∂u. hI are defined in (4. .2) and (5. component is designed as a robust LQ optimal con- troller of the form For a fixed uncertainty vector v. Their drawback is the difficulty of model of 90.4) and the command-actuator dynamics.5) is approximated using Taylor’s series expansion signment design. (5.3). and the in contrary to the classical robustness theory where the desired increment in the trim state.

The adap- hypersonic aircraft is use of an efficient flight con- tive scheme is based on the inverse dynamics of the troller.5) and as presented in Section 5. The common controllability issues controller is required to enable a successful separation faced at the instants when the gain matrices become from the booster rocket. roll rate. mands and sensor feedbacks. and bank angle measurement units. X is to maintain the desired AOA and bank angle to within ±0. The states that are not available for mea- the control laws developed so far meet the flight test surement are estimated using a nonlinear sliding mode requirements. eling uncertainties.109–111 genetic algorithm based ap- tural filter. In the Hyper-X program.5)–(4. 115 are suppressed using a feedforward compensator.14). such an efficient original system.5 Various other efforts to control hypersonic flight are In the design of the Hyper-X flight control scheme.5 degrees during the test flights. Simulation results demonstrate that the con- troller structure proposed in105. One into the control design.112 fuzzy control approaches. the designed controller ap- maintain a desired descent trajectory after the tests plies to a certain flight condition and hence a super- are completed. The controller parameters are gain scheduled with AOA 15 American Institute of Aeronautics and Astronautics .113 and neural namic pressure variations on the AHFV aerodynamics network approaches. It is demonstrated via troller efficient in dealing with the parametric uncer- flight simulations and stability margin analyses that tainties. 106 is successful in gen- eral under a mild assumption about some of the system The basic goal of the flight control system in Hyper- parameters. In the design of the supervisor.and simulations for the nominal flight condition with 6 FLIGHT CONTROL LAWS IN USE V = 15 M. and to pro- switching technique. The control design is aeroelastic vehicles and characteristics of the mod- based on a linear model including rigid-body modes. the pitch rate. However these works which is illustrated in Figure 4. Lead-lag compensators are used to im- flight dynamic models. 000 ft. and to fol- low steering commands from the guidance system to In all the above designs.114. the significant wings (AMW) and twin rudders. is in the development and testing phase. q = 0 deg/sec. To this the control mechanism which generates the aerody- point. presented in the literature. measure of AOA are developed and used. yaw variations in the AHF system due to wide speed ranges rate. classical linear con- pay less attention to the airframe/engine/elastic cou- trol techniques116 are preferred to advanced techniques pling in the hypersonic flight dynamics of air-breathing such as the ones in Section 5.108 optimal and sub-optimal margin requirements avoiding use of a separate struc- control approaches. Several techniques to estimate an accurate approaches will be elaborated in Section 7. a new adaptive sliding mode technique is guided flight.14) instead of the general model (4. Main categories of these ap- prove stability margins. Although a significant number of control schemes for 93 94 The LQ design of is improved in using the triangu. time and/or parameter varying approaches ap- namic surface commands by processing guidance com- pear as alternatives to the gain scheduling schemes. There is still no operational AHFV and the aim of flight tests conducted or planned up to now was to analyze the A recent alternative to the nonlinear controller of 94 is newly developed AHFV components and quantifying the adaptive sliding mode control approach of. h = 110. AHFVs have been designed and tested via simulations lar model (4. Furthermore some specific form vide a controlled descent at the end. The tools to accomplish this task are visor such as a gain scheduling scheme is needed in the aerodynamic control surfaces. observer. They concentrate on applica- second order actuation modes. Elastic mode effects are cir- proaches are model reference adaptive control ap- cumvented in the controller design as a part of the gain proach for linear models. Conventional AOA mea- The design issues in constructing a supervisor for the surement devices can not be used due to aerodynamic complete flight envelop and the use of time varying heating. to achieve and maintain the (almost) singular is circumvented in105. The improvement is illustrated obvious reason for this is that the AHF technology in94 via numerical analysis and simulation results. which are basically order to run the controller during the complete flight symmetric and differential deflection of the all-moving envelop.2) should be taken into account.105. The adaptive structure makes the con- a Boeing/NASA partnership. γ = 0 deg. and dynamics or cer- tion of specific techniques to the existing hypersonic tain filters.5 The control of σ-modification107 is used to preserve stability at laws to perform these tasks are being developed by such instants. and (see Section 3. one of the key technolo- developed for a class of MIMO nonlinear systems in- gies for flight demonstration of the scramjet powered cluding the nonlinear models (4. 106 the AHF dynamics rather than performing a complete In this work. Nonetheless. none of these schemes has imposing nonlinear dynamic inversion techniques104 been completely implemented in a real AHFV. 106 using a design condition during the engine test. The affect of propulsion system and dy- proaches.

first.43 Longitudinal Feedbacks Beside the control laws in use for hypersonic flight AOA/ Load Factor Symmetric AMW tests. Hence. A Lateral/Directional Feedbacks series of test flights with these two testbeds have been planned and a number of them have already been per- Figure 4: Overview of the flight control scheme for X. formed successfully. sions on new combustor designs and the ideas on us- 16 American Institute of Aeronautics and Astronautics . However the con. vanced control schemes such as adaptive neural and robust flight controllers. a for this configuration. accepted in all major AHF research centers of the they remained in the theory and simulation level. although a number of grams the airframe integrated scramjet engine con- control schemes were developed based on the configu.24 niques to fill some of these gaps. 8 Both of these two test aircraft are turbo- Lateral/Directional Command Controller jet powered. cept forms the basis. 115 All these techniques cretized using Tustin (bilinear) transformation for im. gies briefly keeping the focus on control issues. it is also worth to mention the ones applied on Command Longitudinal Deflection subsonic flights of some testbeds with geometries op- Controller timized for hypersonic flight. exit area of the internal nozzle.7. In this section. 117. adaptive engine controllers. Unlike the hypersonic wind tun- 43A designed by Langley and Dryden Research nel and flight tests. some discus- set of engine control effectors were desired to be devel. The non-operated scramjet were performed with rocket next step will be guided AHFs that follow certain pre- power. remotely piloted. so-called advanced airborne data acquisition and con- nonlinear models of the dynamics are used to uncover trol system (AADACS) hardware. since the hypersonic research with of an AHFV operated by an air-breathing engine has X-15 was terminated without performing a AHF test. In addition. In simulations. oped in order to control fuel flow. and the in. and designed for sub- Rudders sonic flights.7.24. This configuration is widely rations used in these programs as stated in Section 2. conventional scribed paths. To this point. cluding controls will be vital in reaching all these fu- Therefore the air-breathing propulsion. neural actuator controllers have been included in these The continuous-time dynamic equations are dis. test flights could not For the current and foreseen AHF technology pro- be performed at all. This goal can be widened to in- surfaces on the tail were used. not been performed yet. 8. experimentation with many ad- Centers of NASA. Then trol techniques developed to enable guided hypersonic we summarize the gaps in the current AHF control flight with X-15s were unique in the supersonic and schemes and propose a set of recently developed tech- hypersonic flight control in many aspects. have been implemented via software installed in the plementation and simulation purposes. flight Hyper-X program. 26 For flight in the atmosphere. subsonic flight tests. One of the main goals of the All the 199 flights of X-15 with or without a dummy current AHF programs is to perform such a flight. ture goals. an essential task is to en- X-30s and HyFLITE vehicles were similar to the ones hance and optimize the propulsion systems developed designed for X-43s —described above.and Mach number. the possible effects of linearization and variations of the system parameters. Simulation and stability anal- ysis results demonstrate that the flight test require- ments are met. hydrogen peroxide thrust rockets on the nose and Development of new technologies in various areas in- wings were used as a reaction augmentation system. we present the ma- teractions and structural constraints brought thereby jor emerging trends and necessities in AHF technolo- were not dealt in the control design. The former aircraft is specially designed to collect subsonic flight data for the AHFV X-43A.5 7 FUTURE TRENDS AND DIRECTIONS FOR NEW CONTROL DESIGNS The controller design issues in the HRE test flights with X-15 were slightly different from those in the As can be deducted from the previous sections. 114. The world. In the NASP and the HySTP. 118 Having the airframe integrated en- control tools planned to be used in scramjet operated gine configuration fixed.4. Outside the atmosphere clude the original SSTO space access task of NASP. The X-43A-LS and the Guidance Differential AMW Hypersonic LoFLYTEr test aircraft are two such actively used System Bank Angle Deflection Vehicle testbeds. 9. The final goal can be thought as to use aerodynamic control systems such as the yaw motion AHFVs efficiently to follow arbitrary paths similar to controlling rudder surfaces on the vertical stabilizers conventional aircraft in a wide speed span including and the pitch and roll motion controlling horizontal hypersonic ranges. inlet/diffuser area ratio.

flight data (see Section 3). Taking the hanced to a more comprehensive model combining significant and fast variations in the AHF system due lateral and directional modes of the aerodynamics to dominant hypersonic regime during the entire flight with the longitudinal mode. An anticipated option to applicability of the design to the entire flight envelop. use of passive and trol surfaces. scheme is required to perform the transitions between ments of a AHF system in parallel with the ongoing these flight conditions. 117 ology and the increase of interest of the academic re- search community in it for the last fifteen years. in spite of the wide use of this method- merical techniques can be found in. and nonlinearities. number of flight conditions at which the dynamics are sented as more dependable alternatives to the sim.11). ing. stability. plicability of the gain scheduling schemes. exit area of the internal noz- The new combustion ideas mentioned above are basi. However. (4. Hence it can be argued that a controller satisfying certain For the parameter varying approaches. the flight con- performance. the significant as well as wide operation ranges and the lack of broad longitudinal-lateral-directional and lateral-directional. However. 55.119–124 x˙ = fp (x. 117 let/diffuser area ratio. completely quantified are generic assumptions for ap- pler conventional models. u. new materials and active cool. manded via an integrated mechanism that processes ties due to aerothermodynamic effects in modeling and the information signals coming from the aerodynamic control as well as increasing the overall flight perfor. outputs. v. 117 Due to aerothermodynamic issues the integrated models. These control effectors should be com- active heat resistant devices will reduce the uncertain.ing hydrocarbons as fuel instead of hydrogen can be In order to implement the control designs based on found in. The conventional supervision ground and flight experiments. In. However. 51 were pre. it is worth mentioning the trade off between Leaving the discussions on use of gain scheduling to comprehensiveness of a model and the complexity the references.2) y = hp (x. The former of these form a fully guided AHF is design of a supervisor to sources can only be suppressed in time via a number handle the variations in the complete flight envelop as of ground and flight tests which have been in progress mentioned at the end of Section 5. Thereby. Each of the con- since the early stages of the AHF research and which trol designs presented in Section 5 applies to a certain will inevitably continue till maturation of the AHF flight condition and therefore. 83.118 for example.5). there The ignorance of coupling effects in modeling as an un. It is worth noting that the use of sumptions in robust control theory and they need to the system integration approach is not limited to mod- be reduced as much as possible in order to build robust eling and control. inputs. ing controller units. To this scheme for AHFVs will be extremely hard if possible. designing a gain scheduling propulsive interactions would be addressed. the aerodynamic and propul- discussed in Section 3. (4.125–128 Further- the integrated longitudinal aeropropulsive/aeroelastic more. approach is used to optimize configuration variables of a scramjet-powered hypersonic cruise missile for max- The most dominant two among the uncertainty sources imizing the air-breathing range. Note from Section 4 that the uncertainties in using them as separate feedbacks to the correspond- the existing models are too large to meet certain as. p) (7.1) variable systems diminish this burden provided the model can be formulated in certain triangular forms similar to (4. zle should be in use as well as the aerodynamic con- cally performance related. v. slow plant parameter variations and sufficient models derived and elaborated in46.11) are respec- would be hard. Ongoing studies and methodology in aerospace applications is gain schedul- discussions on data collection via experiments and nu. and robustness requirements dition specific models (4. mentioned in Sections 3 and 4 are the lack of suffi- cient data on hypersonic flight dynamics and ignorance Another major issue to be addressed in order to per- of coupling effects in modeling.9. the integration high performance flight controllers. 47.54. speed up the data broadening process is performing the design procedure needs to be repeated for a suf- simulations and numerical analysis techniques such as ficient number of flight conditions and a supervisory CFD methods on dependable models of various ele. sive control laws to be applied need to be interac- ing techniques need to be developed and used widely tive. 57–60. Engine control effectors controlling fuel flow.3). a set of techniques recently tively replaced with developed for certain classes of nonlinear and multi- x˙ = Ap (p)x + Bp (p)u + Exp (p)d (7.125–128 we concentrate on use of time of designing/implenting a controller for this model.4. 6. of states. These models may be en. point. and parameter varying approaches as alternatives to More comprehensive models will have larger numbers gain scheduling next. in- in many structural components of the AHFVs. 40. p) 17 American Institute of Aeronautics and Astronautics . and propulsive measurement units together instead of mance. 9. in order to extend the technology if not forever. is still a lack of tests to guarantee global stability certainty source was already discussed in Section 4 and of gain scheduled systems in general.

needed if the output signal can not be measured ac- ear functions and p is a lumped time varying parameter curately. The extended scheme will be x˙ 1 = f1p (x1 . t) + gt (x1 . Since p is X-43A. x2 . A method to exploit the a priori information all major AHF research centers of the world. . ht . and interaction issues. however. systems including (7. . . fit . gp are smooth nonlin. ht . . Ext . the system parameters defining At . . v. eters online will be required in the control design. than performing a complete guided flight delay com- Bt .. Bp . p)u sue is partially circumvented in output feedback de- T signs such as the ones in. Due to lack of a perfect data been performed yet.129–131 These control designs. . 135 in which among x = [ xT1 xT2 · · · xTk ] the plant signals. v. v. x˙ 2 = f2p (x1 . . (7. Aerody- about system parameter in designing controllers for namic/propulsive/structural interactions in this con- time varying systems is presented in. v. gt are smooth func. hp . x2 .5) mators together with nonlinear controllers. . v. Hence adaptive laws to estimate these param.119.5 x˙ = At (t)x + Bt (t)u + Ext (t)d (7. t) . may not be efficient. v. p(t) defines that occurred in the control law development for the the flight condition of the AHFV at time t. t) ple. Lack of an operational AHFV and base.1). and controlling the AHF dynam- ics. nificant progress in various aspects of the hypersonic tions.6).. p) be measured in order to generate the control signal. we have reviewed the literature on char- y = ht (x. As a final issue to be addressed in AHF control design. t) x˙ 2 = f2t (x1 . v. The review demonstrates that despite the sig- where At . xk . fip . . ventional AOA measurement devices can not be used pendance of the system functions in (7.3) due to aerodynamic heating. Ext . recently developed AHF control schemes.132 This method figuration are at a non-negligible level. modeling. . we can treat the de. x3 . 132. Exp . x2 . jet engine configuration is widely accepted as a basis in tures. t)u T x = [ xT1 xT2 · · · xTk ] In this paper. x˙ 1 = f1t (x1 . Bt . . The other 18 American Institute of Aeronautics and Astronautics . only the plant output is required to y = hp (x. to hypersonic speed while some pay attention to all ever. (7. . and unmodeled disturbances components and quantifying the AHF dynamics rather and uncertainties. ft . xk . . The information about these functions that can flight in the last sixty years.. Using linear model based esti- x˙ = ft (x. (7. . p) + gp (x1 . .3) we consider estimation and observer schemes.2).6) 8 CONCLUSIONS x˙ k = fkt (x1 .132–135 Moreover the control addressed in addition to effects of hypersonic speed in scheme of 135 can be extended to a class of nonlinear any affective AHF model or control scheme. Hence the dy- is used to design effective adaptive controllers for linear namic couplings due to these interactions need to be time varying systems in. some estimation may be where Ap . This is- x˙ k = fkp (x1 .4) The estimation techniques to be used. The airframe integrated scram- ploit the a priori information about the variation struc. x3 . how. v. are in general based on the assumption that the major characteristics including the system integration variations are slow in some sense and they do not ex. p) a potential candidate for controlling AHFVs. 136–138 can be used with possible nonlinear AHF controllers. t) (7. p) . for exam- y = ht (x. xk . xk . Such a case is the AOA measurement issue vector defining the flight conditions. Flight of the guidance scheme of the complete flight and the an AHFV operated by an air-breathing engine has not flight condition data base. Majority of the mathematical models and control Time varying and parameter varying methods have al.e. i. x2 . the AHF technology is be used for control design purposes is determined by still in the development and testing phase.5 In hypersonic operation of the X-43A con- basically a vector function of time. u. v. . v. t) acterizing. . ft . fp . (7. v. Newly developed nonlinear observers and estimators119. . fit . Several techniques to as dependance of them on time and further rewrite the estimate an accurate measure of AOA have already model equations as been developed and used in the Hyper-X program. Even with these controllers. v. the difference between the desired (guided) and the focus of the flight tests on analyzing new AHFV actual flight trajectories. need to guarantee no harm to the stability performance of the control scheme. schemes presented in the AHF literature consider only ready been used in designing controllers for various the dynamic characteristics of the AHFV that are due aerospace systems. gt are not expected to be fully plete implementation and experimental testing of the known.

.” [3] McClinton. L. 1191–1198. J.” AIAA Journal of Propulsion and Power . 1969. L. Nov. Rausch.. 2002-5175. D. Sep. Pratt. and Combus- UAV. NY.. D. 17. and Huet. [17] Dugger.. Pahle. Nov. and Hunt. [13] High Mach Number Airbreathing Engines. C. M. L. G. 1997. 1961...... Ox- ford.. compensation of measurement in. Pergamon.” Journal of the Royal Aero- nautical Society. M.” Comptes C.” AIAA Paper 99-4124...” 10th Oxford. 6. ous discussions and for his suggestions in improvement of the paper. New York. F.. J. Daley. “Review of problems in application of AIAA 2001-1910. L. 1. tion. DC. J. and Jorgensen. See also Aircraft+Establishment Library Trans- Congress.. J.main issues to be addressed in designing an effective [9] Curran. “Accurate Automation Cor. Sept... 319–336. H. 17. M.” [19] Ferri.K. No. [23] Tretjakov. [11] Grallert. W. “Synthesis of a FESTIP airbreath- ing TSTO space transportation system. and modeling uncertainties. Journal of Propulsion and Power . 2-8. ACKNOWLEDGEMENT [12] Sakurai.. D. Such approach 2001. 1994. sys.. A. McMinn. J. V. pp. No. G. Flight Test of the X-43A-LS Hypersonic Configuration and Computaion in Flow.” AIAA Paper 2002. chap. “Supersonic combustion: A Russian view. J. poration’s LoFLYTEr Program. “An Analy- [2] Rausch. Vol. J.. Y.” Acta Astro- Doman of the Air Force Research Laboratory for vari. C. in complete flights. of the Bose. 1996. D. and [21] Marguet. Carlton. Wiley... McClinton. trollers based on time varying models. “The NASA rendus de l’ Academie des Sciences. with successively subsonic and supersonic [7] Gibson. T. pp. D. P. M. M. V. 2001. Turbulence. 1997. I. combustion. “Hyper–X program status. 2001.. and Reukauf. R.. fast variations of system parameters. M. and Davidenko. No.” AIAA Paper 2002-3462. C. P. [1] Freeman. Cox. V. C. J.. P. E. 575–597. C. 2002. [6] Heiser. 3-7 range. “French-Russian tential approach that could address all these issues is partnership on hypersonic wide-range ramjets. Cox. L.. Saeks. tem nonlinearities. “Development and [22] Tretyakov. breathing technology. “Flight control laws for NASA’s Hyper–X Third International Symposium on Air Braething En- research vehicle. pp. High Mach Number Airbreathing Engines. 6. Rausch. Pergamon.” J. Kocher. 1968.. D. E. 1967-1993. “Development of the The authors would like to acknowledge Dr... B. C.. [4] Hueter. No. D. 301–338. Lallman. L. I. Modeling. R. David B.. U.. [20] Waltrup.. 105–112. sis of Ramjet Engines Using Supersonic Combustion.. McClinton.” AIAA the use of integrated robust adaptive nonlinear con. R. 5. 835–862. U. has been used to control plants with time varying pa- rameters of the same class as the AHFV models. pp. No. pp. and McClinton. C. AIAA Selected References Reprint Series. U.” AIAA Jour- nal of Propulsion and Power . Washington... mum solution for a fized geometry ramjet. editors.. New York. M. pp. NY.. C. V.. R. 222. U. Sitz. M... “Scramjet Engines: The first AHF controller are variations due to wide speed ranges forty years. Avrashkov. “Mixing controlled supersonic combus- 11th AIAA/AAAF International Conference. V. Levine. C. 1964. Stull. and Mackay.. and Yamamoto. R. 1999. 645. L. L. Vol. Falempin.” 48th International Astronautical 1946.” Proc. tion for a Scramjet.. J. F. S. ISABE 97-7024... 1961. AIAA/NAL/NASD-ISAS International Space Planes and Hypersonic Systems and Technologies Conference.. lation 112. and Anderson. and Crawford. K. “Moteurs thermiques. C.” Annual Review of Fluid Dynamics. H. and Bruno. C. Nov. Apr. [18] Ferri. Stephenson. S. pp. L. [14] Dugger. V. P.. Yamazaki. 68.. J. Cooper.. T. R... H. Kobayasi. in the Mach sion. supersonic combustion. gines. 2002.” XIII ISABE .K. Vol.. 2002. Hyper–X program. 1997. [16] Weber. Vol. Vol. J. Gibson. H.. J. Hypersonic Airbreathing Propul. 3502. Ramjets.. Vol. Oct. pp. [5] Davidson. Golovitchev. 1138–1148.. 6. Reubush. The Study of Supersonic Combus- [8] Neidhoefer. Vol. AIAA.” ONERA TP 656E. G. 1177–1183. R. No. 1973. 1958. Vol.. Selmon. “Hyper–X: Flight validation of hypersonic air- NACA TN 4386. J. Neidhoefer. [15] Roy. “Comparison of hypersonic ram- jet engines with subsonic and supersonic combustion. accuracies. March 1976.. AIAA tion. “Supersonic combustion ramjet(scramjet) en- Martin.” 19 American Institute of Aeronautics and Astronautics . and gine development in the United States. “NASA’s advanced space transportation hypersonic program. and Crawford. 1. 17.. 6. 2001... hypersonic flight experimental vehicle. nautica. D. Experimentation. “Research of an opti- Mehta... Shi- rouzu.. A. A po.. F.. [10] Bouchez. Vol. 40..

R. K. K. 460–466.5. DC. Vol. IAF Paper 93. C. J..” In. “Etude theorique et experimentale d’un Statoreacteur Propulsion. A. P. 2. 1998.” AIAA [40] White. Capriotti. K. “Optimum mission performance R. May 1994. gram Ofiice to the termination of the HySTP scram- tion. Vinogradov. Roudakov. G. O. Paper 72-74. and Chavez. A. and Kopchenov. O.. H. and Anderson. A. Air Force History and Museums Program. demonstration engine and subscale parametric engine G. “An Inte- V.. H. namic Analysis of Hypersonic Vehicles. K. and Lovell. [32] Voland. R. A. K. 1992.” Fizika Goreniya i Vzryva.” AIAA Journal of Guid- “Scramjet Integration on Hypersonic Airplane Con. J. K.. considerations for the airframe-integrated scramjet.. A. L. 1999.. March 1997. Van Nostrand Rein- ternational Council of the Aeronautical Sciences. and Mackley.. 17. Vol. 1991. Iliff. Jan. tests. 1308–1319. P. 1973. S. and Kobyzhskii. Marguet. Semenov.” AIAA Paper 95-6055. “A historty of the gram... L. J. model with dynamic analysis. R. D. F. 1992.. and Schmidt.. V. and [44] Schmidt.. Y. I: From Max Valier to Project PRIME AFB. Dec. The National Aero-Space Plane Program (1983–1995). Dec. “Design and modeling issues for an annular chanel.. P. T. J.” AIAA Paper 99-4848. “Flight testing an ax. H. R. 1991. At the Edge of Space: The sonic X-plane development of technologies and capa- X-15 Flight Program. Fuzzy. R. J.44th.. G. 6. (1964–1986). No.. G... [49] Schmidt. “Comperative flow path analysis and sonic aeropropulsive/aeroelastic vehicles. Vol.” AIAA Journal of Aircraft.. 158–164.1–VIII–1. D. April 1995. Nov. “CIAM/NASA Mach 6.. [38] Chase. editor. Oct. Smithsonian Institution Press. D.. Y. tegration challenge for the 1990s. W.. “An- NASA TM X-2895. D. [27] Andrews.. E. editor. and Tang. New York. 20 American Institute of Aeronautics and Astronautics ... Roudakov.4. P. V. A. F. June 1992. L.Proc.” AIAA Paper 96-4571... pp. T. pp. Y.. Vol. R. G. and Dynamics. Bolling AFB. DC. per 95-6051. June 1998. and Guy.. Bolling tion. “Dynamics and control of hyper- Kopchenov. Semenov. Menousek. 24–29. [37] Schweikart. Schickhman.. F. [36] Rogers. jet performance demonstration program.” NASA ARC [33] Henry. Joint Meeting of the Italian and Spanish Sec. K. “Integrated control of hyper- sonic vehicles. NASA’s Hypersonic Research Engine Project. Auslander. Vol. [31] McClinton. Small. craft and Hypersonic Vehicles. and Penland.. seums Program. pp. Semenov.. and Schmidt.. R.. chap. scramjet flight test engine at a Mach number of 6. D. 1986. Vol... W. Zhadan. R. NASP program from the formation of the Joint Pro- [25] Hallion. Flight. 34. tion of hydrogen combustion and supersonic cooling in [42] McRuer. V. Vol. flight vehicles. Vol. pp. A. and Rockets. [45] Schmidt. bilities for 21st Century access to space. A.” AIAA Paper 98-2506.” In. “Design 92-2. 1993. isymmetric scramjet: Russian recent advances. [24] Hallion.. “NASP concept [29] Albegov. 729–734. P. 1992. A. and Anderson. Novelli. ture Aerospace Technology in Service to the Alliance.. pp. No. 5057.” AIAA Paper 91- S. 357–465. W. H. Handbook of Intelligent Control: Neu- [28] Contensou... 1999. II: From Scramjet to National Aero. formance and design sensitivities of air-breathing hy- sonic combustion research at Langley. M. B. (1924–1967). American Control Conference. 1998. I. and Rock. C. NY.. pp.. 1991. cepts. 1995. H. A. 5. 1998. No. M. [43] Schmidt. pp. “Super. Bolling AFB.. ral. D. A. R. alytical aeropropulsive/aeroelastic hypersonic-vehicle [34] Weidner. [48] Schmidt. R. 6.” AGARD Fu- Washington. pp. V. 1977.” AIAA pa- design assessment of an axisymmetric hydrogen fueled per 92-4326. 1993. and Trippensee. 1996. “Experimental investiga. K. tion. The Hypersonic Revolu. Control. D... Volume III: The Quest for the Orbital Jet: VIII–1.. A. 1972. Smart.. D.” AIAA paper 93-5091.. launch vehicles. “Dynamics and control of hypersonic vehicles The in- ternational Astronautical Federation. J. DC. and Fourt. “Review of April 1997. V. Mamich.. Vol. DC. [30] Roudakov.” [47] Chavez.” Proc. ance. ICAS hold. “Mission per- [35] Northam.. June–July 1993. [41] Voland.485. G. S.4. 12–1–12–10. [39] Hicks.” AIAA Pa- Space Plane.5 scramjet flight and grated Analytical Aeropropulsive/Aeroelastic for Dy- ground test. The Hypersonic Revolu- tions of the Combustion Institute. “Experimental supersonic combustion research and multivariable flight guidance for airbreathing at NASA Langley..” AIAA Journal of Spacecraft 86-0159. V. R. D. and Huet.” AIAA Journal of Guidance. 27. The Hypersonic Revolu. Aug. Air Force History and Mu. P. [46] Chavez. and Paper 93-2323.” AIAA Paper personic launch vehicles.. D. K. and Thermal Control of Advanced Air- a combustion mixte (domaine de vol Mach 3/7). E. “NASA hyper- [26] Thompson. Control. 14. and Adaptive Approaches. integrated airframe/propulsion control of hypersonic No. Bowers. Air Force History and Museums Pro.. C. editor.

C. Giffith. L... 38. “Aerodynamic Database gram PREPHA. “Hypersonic Mach Number and Real 21 American Institute of Aeronautics and Astronautics . 1999. to the Puzzle. S. ISABE Paper 99- cle. F. Engelund. Vol. “Pro- [55] Engelund. Engine Ground Testing for Flight Risk Reduction.. “Hyper-X Stage Separation Wind-Tunnel Test Program. A. 2001. N. and Andrews Jr. A. Vol. “Hyper-X Flight pp. and Frendi. Rock. D. Best. “Experimental Study of frame Integration Issues and Aerodynamic Database Scramjet Modules. S. K. [66] Sancho. [69] Yatsuyanagi... E. N.. Jr.. 32–34. “Status of 2001.. [68] Novelli. 2001. W. I. pp. pp. Lanshin. 22. Vol. 6. and No. A. T.. C. and DiFulvio. No. Sept. sonic vehicle.. Guidance. 87–95.” Proc. C. gram Overview: The French Hypersonic Research Pro- C. V. periment. “Computational Fluid Dynamics Prediction 99-7091... S. H. D. Bittner. R. W. posium on Air Breathing Engines. 6. T. Holland. J. Whitehead.. B. G.. No. D. D. AIAA. G. J... Holland... Szema.” AIAA Journal of Spacecraft and Rockets. No. D. T. Dilley. Advances in Hypersonics..” AIAA Jour.. No. A. Holland. T. 1999. S. 811–819. W. Progress in Astronautics and Flight-Test Experiments.” AIAA Journal of Spacecraft Aeronautics. 2001. Robinson. “Hyper–X Stage [50] Schmidt. Control. [59] Cockrell Jr. Vol.. ed- nal of Spacecraft and Rockets. 38. [71] Anderson. itors. R. [67] Falempin. R. pp. K. W. and Rockets. DC.. E. M. Nov.. of Hyper-X Stage Separation Aerodynamics. T. A. W. The X-43A Airframe-Integrated Scramjet Propulsion Scramjet Propulsion. 38. sis of elastic high-speed vehicles. pp. [70] Bertin. V. 1. pp. D.” AIAA Journal of Spacecraft and Rockets. J. and Johnson. 38. the 7th AIAA International Development for the Hyper-X Airframe-Integrated Space Planes and Hypersonic Systems and Technolo- Scramjet Propulsion Experiments. Spacecraft and Rockets. D. “Integrated Aeropropulsive Computational Fluid [72] Bertin... Scramjet Research at NAL. and Dynamics.” AIAA Journal of Spacecraft and Rockets. 836–843. Sept. C. and Symposium on Air Breathing Engines.” AIAA Paper 99-4814.” Proc... L.... [51] Chavez. [53] Engelund. Voland. S. and Weber. K.” AIAA Aerospace America. “Uncertainty Demonstration. Boston. Jan. M.” AIAA Journal of [63] Sosounov. P.” ISOABE Paper 99-7215. “Hyper-X Research Vehicle Experimental Aerodynamics Test Program Overview. D. S. “Hyper-X Aerodynamics: [65] Curran. and Serre.. J.. Ruf.” Proc. 1999. K. Vol. 21.. D. 6. and Vinodragov.. V. 1999. Washington. and Bittner. K. Nov. 1989.. D. J. L. and Chap- man.. 820–827. E. D. Sept. M. DC. 844–852. E. 6. pp. J. 1994. the 20th Interna- tional Symposium on Space Technology and Science. R. Starov. “NASA’s Hyper-X Scramjet Engine 12th International Symposium on Air Breathing En- Ground Test Program. A. V.. and Engelund. C. Jr. Semenov. C.. Joint ONERA–DLR Research Project on High Speed Airbreathing Propulsion..” AIAA Journal of gies Conference.and Dynamics. the Huebner. pp.. W. Woods. [62] Faulkner. No. and Ro- 1999. 1. P. Nov.” AIAA Journal of Spacecraft and Rockets. Dilley.. 2001. Hypersonic and High Tem- R. Martin.” AIAA Paper 99- modeling for multivariable-control robustness analy. and Application. Vol. Colin.. AIAA 2001-1802. C. the 14th International [57] Buning. 1987. Wong. N. and Bittner. C.” AIAA Journal of Guidance. April 2001. Jentink. J. No. “Fitting Aerodynamics and Propulsion in [60] Huebner. J. Vol. T. Witte.” energy-state analysis on a generic air-breathing hyper. C. 803–810. McGraw-Hill. D. 1992. “Propulsion System Air. 1157– [61] Reubush. N. C. J. F... O. “Two Generations of Russian Scram [52] Rock. Nov. P. J.. A. C.. B. J. W. “Hydro- and Dynamics. A. and Ballmann.. 801–802. Roudakov. 1997... 1996. “Use of Separation—Simulation Development and Results..” Proc. 2001. [73] Johnson.. J. R. Y. D. Rogers.” ISABE 99-7214. 6... J. V. 20. Jan. MA. H. May 1996. [54] Engelund. E. Y.. 4922. No. “The French PROMETHEE Program: Main Goals and Status in [56] Woods. editors.. B. A. F. A. 1998. J. I.. K.. W. Nov. 38. and Strovers. 6.” Proc.. N.. [64] Goldfeld. 828–835. Dynamics Methodology for the Hyper-X Flight Ex- AIAA. 1995.. gines. S. 38. Nov. Washington.. Sept.. No. Control. Nov.. L. and Chinzei. D. D. Cockrell Jr. E. No. Nestoulia.. 2001.. and Murthy. 7052. R. [74] Maus. Nov. pp. [58] Holland. Cockrell Jr. 1164. 1999. G. J. Birkh¨auser. “JAPHAR—A Vol.. M. R. 38. D. J... W. 6. pp. carbon Scramjet Propulsion System Development. Vol. C... Hypersonic Aerothermodynamics. G.. Kopchenov. The Oriel Programme. and Schmidt. E. D. ISABE Paper Pao. Periaux. S. 6. Bose. Vol. perature Gas Dynamics. 1999. and Hermann.. Nov. and Flying Test Beds. W. mankov. 71–76. E. 2001... V.... and Koschel. R. the 14th International Sym- Development for the Hyper-X Flight Research Vehi. V. Sept. E. pp. Paper 96-a-2-10. 1999.

Vol. H. 1. J. “NASP Aerodynamics. McMinn. [91] Gregory. and Pittman. CR-1065.. J. Vol. 1992. and Dynamics. 1. J. and Stengel.” “Investigation of Piloting Aids for Manual Control of NASA TM-102610. [81] Walton.. L. of the X-43 Hypersonic Flight Vehicle. H. “CFD-Based Aeroelastic Analysis ness. L.. January 1991. hicle. Jan. 0379. “Design of ro- J... and Calise. Pinckney. 1998. R.. “Integrated ics. 1987.. “An Aero Dynamic Analysis of Several Hyper. S. Upper Saddle River. 1994. K. pp.” NASA TM-4562. 1989.” AIAA Journal of Aircraft.. D. 577–585. and Dynamics. “Aeroelasticity of a Generic Hy. 1989. J. R. K.” AIAA Journal of Guidance.. “Multidisci. pp. and McMinn. T. and Shaughnessy. 20. [90] Shaughnessy. 21.. E. I. A. J. A.. No. Z. K. Nov. personic Aerodynamic Model Example (GHAME) plinary Aeroelastic Analysis of a Generic Hypersonic for computer simulation. Dinlon. [94] Wang. Addison-Wesley.. L. and Doyle. C. D. “Design of High Prformance Ram. Paper 99-0765.” AIAA Paper 89-2463. and Iliff. Vol.. W. 1.” AIAA Journal of Guidance. Jan.” Proc. D.. C. D. “Development of a Discrete-Time Aerodynamic [99] Zhou. [98] Maciejowski. and Schmidt. J. 1. E. 1984. A.... Vol.. bust control systems for a hypersonic aircraft. K. F. D. Edwards. 18. A. namic Pressure. 1977. “Hypersonic vehicle [77] Chalk..... and Dynamics.” NASA Technical Paper 3525. J. [79] Penland.” NASA TM. Vol. C. [89] Bilimoria.” AIAA Paper 99-4123. jet or Scramjet Powered Vehicles. NASP JPO. [100] Bowers. 1990..Gas Effects on Space Shuttle Orbiter Aerodynam. bustness of Linear-Time-Invariant Control Systems. No. A. and Petersen. and Stengel. P. No. M. [86] Thuruthimattam. C. Wright-Patterson AFB. and Stengel. “Generic Hy- [85] Gupta. 58–63. “Flying Qualities Criteria Review. F. Pototzky. 41. 1989... Dec. L. H. [92] Brauer. L. Vol. Jonckheere. 132–141. 716–723. D. “Performance Sensitivity of Hy. July 2000. mary Document. B. Control. S. 1993.” AIAA Journal AIAA Paper 89-5013. Cruz. F. CA. Cornick. J. K. D. 42–48. AIAA Journal of Aircraft.” personic vehicle model. and Sparks. L. Jan.. L.. Vol. F. R.” AIAA Control .. Powell.” AIAA sonic Research Airplane Concepts from M=0. NY. and Kelley. 1994.. 23. “Robust nonlin- [80] Whitehead... Optimal Control and Estimation. pp. sign. J. 1998. 22 American Institute of Aeronautics and Astronautics . 32. pp. [97] Ray. and Stengel. NJ. August 1999. “Uncertainty [96] Ray. L.. and Dalzell. Hypersonic Flight Vehicles. “Multi. personic winged cone. Jr. J.” Journal of Guidance. D. March 1995. Nov. C. 21.” NASP synthesis. 1. R. Hypersonic Maneuvers. Approach to the Analysis of Control System Robust- T. 15. and Gupta. Buffington.. K. “Impact of Aero-Propulsive-Elastic Interactions [102] Young.. C. ance. K. 2333–2337. 11. K. M. Bach. A. “Stochastic Ro- modeling and fixed-order controller design for a hy.. New York. Dover.. 1995. 2002. Dryden. [95] Stengel. D. M. 2. “Hypersonic Ve- [76] Raney. [78] McRuer. J.S. model and control law development using H∞ and µ- Assesment and Recommendations for NASP. S..2 to 6. K. Multivariable Feedback De- [84] Cowan. A. 1989. 82–87. “µ Controllers: and McNamara. H. P. A. No. IEEE Transactions on Automatic Control . 229–236. hicle Simulation Model: Winged-Cone Configuration. and Dynamics. 34– [87] Raney. J. Control. trol. S. pp. “A Monte Carlo [83] Gupta.. 2001.... R. 1997. M. 1989. L. D. Control. 1. [101] Bushcek. 1997. E. T.. Control.. “Eigenstructure vs constrained H∞ design for hy- cle..” AIAA Paper pp. Development of the Equations of Motion for Elastic No. Phillips. L. Voelker. “Considera.” AIAA Paper 93-5028. “Controller Design with Mixed on Longitudinal Flight Dynamics of a Hypersonic Ve. T. NASA Ames- Vehicle. McMinn. Doyle. Chowdhry.” ear control of a hypersonic aircraft. R. pp.” AIAA Journal of Guid- [75] Schindel.0. and Stevenson. 1994. No. October 1995. No. S.” AIAA Paper 2002-1209.. “Capabilities and Applications of the Program to Op- tions for the Development of NASP Flying Qualities timize Simulated Trajectories(POST)—Program Sum- Specifications. I. [82] Bushcek. 4. 1999.. [103] Lohsoonthorn. Q.. L. Vol.. January 1993. of Guidance.. Con- personic Vehicle. 36. No.” AIAA paper 89. 1978. American Control Conference.. Air Force.. pp. Mixed and Fixed. and Calise. and Hahn. Jan... M. and Myers. Prentice-Hall. 73–81. Friedmann. Vol. P. R... J. 20. personic Vehicles to Change Angle of Attack and Dy.” WL-TR-92-3042. NY. pp.. Arena. I. L. R. pp. D. Unceratinties. F. Vol. R. K.” AIAA Journal of Guidance. No. Loop Aeroservoelastic Control of a Hypersonic Vehi. 01-0712.” AIAA Journal of Spacecraft and Rockets. J. R. and Dynamics. T. G. Essentials of Robust Model for CFD-Based Aeroelastic Analysis.. S. R. Jr. K. J.” Automatica.. New York. H. [93] Marrison. R. 29. U..” NASA CR-2770. and Person. K. [88] Lind.

. E. Jan. “Multidisciplinary optimization pp. Proc. pp. “Missile Autopi- [115] Saeks. G. 4. 2002. A. Nov. Dec 1997. sign: A VSTOL Flight Application. G. K. 2001. Q. L. H..” IEEE hypersonic vehicle. Neidhoefer. Control. 2000. Wiley. Control. 5. 1998. 1996. NY. Control. 898–907. Adaptive. and Lendaris. [118] Bowcutt. P. and Pap. 1993. Cox. “Sub-optimal Vol. 8.” AIAA Journal of Guidance. A. 4.. New York. E.” AIAA Journal [104] Isidori. C. [130] Zhu.. P. lot Design Using a New Linear Time-Varying Control “Neural control of the LoFLYTEr aircraft. Robust Control of Nonlinear Unceratin [108] Mooij. 648–658. and Mirmirani. H∞ Aerospace Control De- port. 4688–4693. and Ioan- periodic control and regulation in constrained hyper.. 2428–2433. tromagnetic Actuation. P. M. AIAA Paper 2001-4168. and Athans. M. P. spondek. Con. Apkarian. No.. C. M.” P. New York. H. “Neural adaptive control of the LoFLYTEr . H. No. Vol. Vol.. [131] Biannic. “Fuel-optimal [124] Fidan.. R. pp. [122] Qu.. S. Boussalis. S. ence. 1186–1190.. D. NY. D. [128] Rugh. hypersonic flight vehicle. Vol. editors. Vol.” Proc. American Control Conference. March 2001. editor... Conference on Systems. M.. No.” AIAA [123] Isidori. “A Switching Controller for Multivariable sonic flight. W. Kanellakopoulos. Man..” Automatica. 1997. J. pp. L. ing: Potential Hazards and Possible Remedies. [105] Xu. pp. Oct. J. pp. [111] Naidu. of Propulsion and Power . Lamnabhi-Lagarrigue. American Control Confer. 17. Vol. Varying Systems: Control and Adaptation. [110] Chuang. Lewis.” Proc. Prentice- 2001. and Ioannou. 24. [125] Hyde. B. erence adaptive control for hypersonic aircraft. 6. 41st IEEE Conference on Decision and Control .. A. D.. Vol. and Dynamics. pp. and evaluation requirements for 1997. 35. 1184–1190. “Analysis of “Unified approach to H2 and H∞ optimal control of a Gain Scheduled Control for Nonlinear Plants. and Maurice. Part 2. 2730– 2735.. “Application of [127] Shamma. “Adaptive sliding mode control of a linear Unceratin Systems.. Springer-Verlag. 10. Robust Adaptive Hall. F. [116] Ogata. Neidhoefer. 2. and Mickle.... A.” Dynamics. of airbreathing hypersonic vehicles. and Speyer. A. V. 1998.. test.” Advances in Astronautical [121] Marino. H. and Lin. P. 23 American Institute of Aeronautics and Astronautics . 36. R. 2000. P.. Wiley. and Athans.” IEEE IFSA World Congress and 20th NAFIPS International Control Systems Magazine.” AIAA Journal of Guidance. [132] Tsakalis.-Feb. 20. J.” Control. and optimal periodic solutions for hypersonic trans. 315–323. and Morimoto. 1998. 2001. pp. Design — Geometric. Vol. “Numerical investigation of model ref. C. June 1999. October control system for hypersonic transporter. Conference. P. Vol. pp. J. and Jacobs. Upper Saddle River. L. Control. [109] Dewell. No. 24. S. and Garrard.. “Parameter Varying Control of a High- Prentice-Hall. J. Prentice- [107] Ioannou.. Saeks. “Research on [113] Zhou. C. NJ.” Proc. 2002. Z. S. L. and LTI Systems with Known and Unknown Parameters. Nonlinear and Adaptive Control Design. pp... future airbreathing hypersonic systems. and Dynamics. 1995. 6. Proc. B. and Buffington. 1997. 101–107.” AIAA Journal of Guidance. Kosmatopoulos. Transactions on Automatic Control . UK.. June 1995. 2737–2741. 1361–1365. W. and Tomei. J... J. 1.Control.” Proc. Springer-Verlag. [106] Xu. Z.. pp. No. J. pp. and Sun. July 2001. Vol. “Linear Time-Varying Ap- proach to Satellite Attitude Control Using Only Elec- [114] Cox. No. J. Vol. S. pp. S. J. Stabilization of Non- and Ioannou.” AIAA Jour. [129] Wi´sniewski. R. nou. Prentice-Hall. A. [126] Shamma. Performance Aircraft. R. 4. J. Mirmirani. ference on Decision and Control . and Dynamics. Hall. 2. Linear Time nal of Propulsion and Power . August 1990. NJ. 23. Vol. NJ.. Upper Saddle River.. Springer-Verlag.. [117] Marren. London. Springer-Verlag. July-Aug. No. Englewood Cliffs.” Proc. March-April perimentation. Banda. 1995. 2000 . Sep. “Ex... 20. 4... Nonlinear Control in the Year No. J. No. Nov.. F. Vol. “Robust adaptive [119] Krsti´c. July 2001. 1947–1962. K. 2001. 2913–2917. Dynamics. 1401–1425. and Kokotovi´c. 20. I.. W. Modern Control Engineering. K. June 1992. “Gain Schedul- genetic algorithms to hypersonic flight control. Systems.. H.. and IEEE Int. H. M... and Deng. 1995. netics. J.. “Fuzzy logic based flight Gain Sceduling. Vol. M.. American Control Conference. K. M. and Robust. R.” AIAA Journal of Guidance. pp. and Cyber. A. 3112–3117. and Dynamics.. Nonlinear Control Systems. R. S. 2001. P. L. 17. Leung. 923–932. 2001. and Shamma. sliding control for a class of MIMO nonlinear systems. 4. [120] Krsti´c. 4. M. C. Technique. Control . and Re- Journal of Guidance. M. No. pp... 1997. pp. [112] Austin... Nonlinear Control Sciences 108.

2000... No. and Kokotovi´c. 24 American Institute of Aeronautics and Astronautics . [138] Arcak. 507–523. Y.. 1290–1300. Vol. 37. December 2001. Springer- Verlag. Vol. No. pp. 74. 12. 637–662.” Automat- ica. New Directions in Nonlinear Observer Design. pp. [136] Nijmeijer.” Int. Vol. S. M. [134] Limanond. on Automatic Control . A. S. 5. T. P. “Constructive nonlinear control: A historical perspective. 2003. P. B. “Nonlinear ob- servers: A circle criterion design and robustness anal- ysis. Vol.[133] Limanond. and Ioannou. [137] Kokotovi´c. K. H.. P. K. on Automatic Control . “Backstepping Control of Linear Time Varying Sys- tems with Known and Unknown Parameters. 37. “Model refer- ence adaptive and nonadaptive control of linear time- varying plants. pp. S. 45. “Adaptive and nonadaptive pole-placement control of multivariable linear time-varying plants. of Control .. pp. M. [135] Zhang. 2001.. I.. editors.” to ap- pear in IEEE Trans. Fidan. and Tsakalis. 1923–1930. and Tsakalis. and Fossen. 1999.” IEEE Trans. and Arcak. J. 2001..” Automatica.