Sunteți pe pagina 1din 17

AIAA 2012-0252

50th AIAA Aerospace Sciences Meeting including the New Horizons Forum and Aerospace Exposition
09 - 12 January 2012, Nashville, Tennessee

HIFiRE 6: An Adaptive Flight Control


Experiment
Michael A. Bolender

Jerey T. Staines

Downloaded by Alison Stewart on September 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-252

Douglas J. Dolvin
Air Force Research Laboratory, Wright-Patterson AFB, OH 45433
The Hypersonic International Flight Research Program (HIFiRE) is an
international collaboration between the U.S. Air Force Research Laboratory and the Defence Science Technology Organisation in Australia. The
HIFiRE program is investigating basic hypersonic phenomena and experimental technologies critical to the development of next generation aerospace
vehicles and the understanding of the environments in which they must operate. Further this program will advance the state-of-the-art in a number
of disciplines by progressively evaluating component technologies having increasing levels of complexity and likeness to those of expected operational
systems. The approach is to utilize sounding rockets in order to mature
a low-cost approach to hypersonic flight test. To this end, HIFiRE 6 will
evaluate the tracking performance of an adaptive flight control system on a
representative hypersonic vehicle that is executing a set of predefined maneuvers. This paper describes the current HIFiRE 6 mission and vehicle
system to execute the experiment, along with a description of the adaptive
control law and simulation results.

I.

Introduction

In the past decade, both NASA and the Air Force have had significant programs that
have focused on the flight testing of air-breathing hypersonic vehicles as a means of advancing the state-of-the-art in scramjet engine technology. The successful flight tests of the
X-43A and the X-51 have shown that significant progress has been made in the engine technology necessary to realize scramjet-powered aircraft. Concurrent to the X-51 program, the
Hypersonic International Flight Research (HIFiRE) Program, a collaboration between the
Air Force Research Laboratory (AFRL) in the United States and the Defence Science Technology Organisation (DSTO) in Australia, has been underway. The purpose of the HIFiRE
Program is to provide a means, via flight test, to collect high resolution data which cannot

Aerospace Engineer. Associate Fellow AIAA.


Aerospace Engineer. Member AIAA.

Principal Aerospace Engineer. Senior Member AIAA.

This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States.

Downloaded by Alison Stewart on September 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-252

be otherwise captured in a ground test facility. There are currently eight research flights
that comprise the HIFiRE program, each of which addresses first and foremost a primary
scientific objective, along with several secondary and tertiary experiments. The primary objectives encompass a wide spectrum of disciplinary-specific research problems ranging from
the determination of boundary layer transition on-set, evaluation of combustion kinetics,
radical farming, inlet starting, and guidance and control.
Within the HIFiRE program, there are two flights devoted to solely to the guidance and
control of hypersonic vehicles: HIFiRE 4 and HIFiRE 6. The primary experiment of HIFiRE
4 (see Reference [1] for more details) is to collect flight data on the aerodynamics, stability,
and control of an advanced waverider configuration. HIFiRE 4 will launch two identical
vehicles on a ballistic, exo-atmospheric trajectory, with one vehicle to be flown by DSTO
and the other by Boeing. The research vehicles separate from the booster after second stage
burn-out. Once at apogee, the vehicles re-orient themselves using cold-gas thrusters. As
they enter the test window, the vehicles execute a controlled pull-up from -70 to -50 degrees
flight-path angle. HIFiRE 4 also serves as a proof-of-concept and risk-reduction of the flight
control and the trajectory for HIFiRE 8, which will integrate an advanced three-dimensional
scramjet engine with the HIFiRE 4 airframe to flight test the combined aero-propulsive
configuration.
HIFiRE 6, the subject of this paper, takes a rather dierent approach than that employed
in HIFiRE 4. First, the primary experiment is the evaluation of an adaptive control law.
We have also decided to design an advanced configuration that is representative of a class of
vehicles being studied by the Air Force for a variety of missions. In particular, the vehicle
being flown for this flight experiment will have a highly integrated, inward-turning inlet.
Instead of the ballistic trajectory that is flown on HIFiRE 4, the vehicle will be launched
on a more suppressed trajectory that directly places the aircraft in the test window and
is more representative of a cruise condition. In order to remove the additional uncertainties
due to combustion process, we are focusing on the eects of the uncertainty due to the
inlet aerodynamics across the its Mach number operability range; therefore, this flight, like
HIFiRE 4 is also un-powered. However, unlike HIFiRE 4, there are currently no plans to
perform a follow-on flight with an operating scramjet engine. The aim of this paper is to
provide an overview of the flight research project and to discuss the preliminary adaptive
control law and its performance in a three degree-of-freedom simulation (i.e., longitudinal,
rigid-body dynamics) of the research vehicle flying a nominal trajectory.

II.

Mission and Trajectory

To obtain the required test conditions for the flight research vehicle (FLIRV) using the
Terrier-Oriole booster launch system (LS), the integrated launch vehicle (LV) trajectory will
dier significantly from those used for typical sounding rocket applications. The LV will be
spin-stabilized once the vehicle system exits the launch rail in order to satisfy range safety
requirements. After the first booster stage separates, the second stage will be passively despun, and then rolled to zero degrees bank angle. During second stage burn, the vehicle will
be guided along its trajectory using the control surfaces on the FLIRV. Once the second stage
burns out, the FLIRV will separate, and continue to coast to the test window (Figure 1).
At that point, the shroud will separate from the FLIRV. The FLIRV will then glide while

105

M5

M4

M7

M6

750psf

100
95

1000psf
90

Altitude, ft

Downloaded by Alison Stewart on September 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-252

holding the dynamic pressure constant throughout the test window. From launch to impact,
the mission timeline extends just over 600 seconds, of which approximately 90 seconds are
spent within the test window. A timeline describing the sequence of flight events is shown
in Figure 2. While in the test window, a turn maneuver will be executed to capture the
vehicle dynamic response and the stability and control parameters for the airframe. Modeling
and simulation will be used to determine the specific maneuvers to be used during the
experiment in order to optimally evaluate the performance of the adaptive flight control
system. Following the vehicles exit from the test window, additional data will be collected
along the trajectory. The intent is to fly the vehicle as long as possible until either the
vehicle telemetry is lost, the vehicle impacts the ocean, or the vehicle disintegrates due to
thermal breach and subsequent structural failure; however, the time-of-flight outside of the
test window is not a requirement that drives the structural design and material selection.

85
80
2000psf
TestWindow

75
70
65
60
55
3500

4000

4500

5000

5500

6000

6500

7000

7500

Velocity, ft/sec

Figure 1. HIFiRE 6 Test Window

III.

Vehicle Configuration

The notional flight research vehicle (FLIRV) is shown in Figure 3. The material composition of the FLIRV is still being determined, but it is known that the vehicle will have
a 3D inward-turning, stream-traced inlet and a flowpath representative of a hypersonic vehicle, although this vehicle will be un-powered. It is also desired that the vehicle have the
characteristics of an aircraft. This is evident in the winged configuration shown in Figure 3. The launch vehicle (LV) consists of the FLIRV and launch systems, and any newly
constructed ground support equipment (GSE) that is used to service, test, or support the
LV. The current baseline configuration of the LV is shown in Figure 4, consisting of the
3

HIFiRE-6 Trajectory to End of Flight


Start of Test
Qbar= 1000 psf
Mach= 7.34

120000

100000

HIFiRE-6
Launch
Terrier Burnout

Altitude (ft)

80000

Oriole Ignition
Oriole Burnout
Start Test
End Test

60000

40000

20000

Downloaded by Alison Stewart on September 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-252

0
0

100

200

300

400

500

600

700

Time (sec)

HF6 Trajectory Plot for ASM Paper Jan 2012.xlsx | H Complete

Figure 2. HIFiRE 6 Mission Timeline

launch rail system, and the FLIRV and launch systems integrated via the FLIRV support
module. Flight research vehicle integration, the launch system (LS), test and evaluation of
the LV, and most of the range operations are the responsibility of the NAVSEA detachment
located at White Sands Missile Range (WSMR). The baseline LS is comprised of a two-stage
sounding rocket (Terrier-Oriole), integrated with a vehicle instrumentation module (VIM)
and vehicle service module (VSM) for vehicle management and data acquisition during the
boost phase. Trajectory data and timelines presented in this paper are based upon nominal
flight performance of the baseline research rocket configuration. The primary function of
the LS is to provide the necessary trajectory to support the research objectives of the flight
experiment in addition to providing data telemetry, power, and some payload management
functions. The FLIRV system (FS) consists of the experimental hardware and the necessary support systems for accomplishing the research objectives: a deployable shroud, FLIRV
support module, and in this case, the FLIRV.

IV.

Adaptive Flight Control

Air-breathing hypersonic vehicles present unique control challenges due to a) the integration of the propulsion system with the airframe, and b) the presence of uncertainties in the
aero-propulsive forces. The integration of the propulsion system into the airframe results in
a system whose aerodynamics (lift, drag, and moments) change as a result of changing the
control input (throttle) to the engine. For example, in the longitudinal axis, a change in
angle-of-attack changes the net thrust output of the engine due to a change in the amount of
air-mass captured, assuming that the throttle position is held fixed. Naturally, this change
in engine performance couples back into the aerodynamics, thus changing the aerodynamic
forces and moments. In the language of the flight control engineer, this means that the A
matrix, and therefore vehicles open-loop stability and control, is a function of the engine
4

Downloaded by Alison Stewart on September 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-252

Figure 3. HIFiRE 6 Flight Research Vehicle

control input.
Within the realm of conventional subsonic aircraft, the aerodynamics and stability and
control characteristics are well characterized through a combination of computational fluid
dynamics (CFD) and ground testing. Ground testing and CFD serve to reduce the uncertainty in the vehicle dynamics before flight test. During flight test, the vehicles envelope
is incrementally expanded, and the aerodynamics and stability and control are fully characterized, along with the performance of the control system and autopilot, before the aircraft
ocially enters operation. However, due to the limitations in the size of current ground test
facilities, we are unable to perform aero-propulsion ground tests on integrated air-breathing
hypersonic configurations larger than the scale of X-51 or X-43 at their design flight conditions. However, recent experience with both X-43A has shown that ground test will not
necessarily resolve the aerodynamics of hypersonic configurations to sucient accuracy to
model the dynamic response and therefore design a control system to guarantee stable and
controlled flight.2 In the case of the X-43A Mach 7 flight test, the control system gave
unexpected responses in flight; however, the excursions seen during flight were fortunately
not too severe, so the control law was robust enough to handle the anomalies and the vehicle was not lost. As the dependence on modeling and simulation increases in the future,
it becomes important to recognize that the amount of uncertainty in the aerodynamic and
propulsive forces and moments will be much more significant when compared to current
vehicles. As a result there will be inherently more risk in the vehicle system in terms of
both its performance and stability and control characteristics. For air-breathing hypersonic
vehicles larger than the X-43 or X-51, the first flight of a particular vehicle will be also
5

Downloaded by Alison Stewart on September 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-252

(3D) 06 May 2011

Figure 4. HIFiRE 6 Integrated with Terrier-Oriole Launch Vehicle

Downloaded by Alison Stewart on September 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-252

be its first test of the integrated aero-propulsion system. In contrast, the large turbofans
that power modern, long range commercial passenger aircraft are thoroughly tested in flight
on surrogate aircraft as part of their flight certification. With this inability to ground test
large-scale integrated air-breathing hypersonic vehicles, in the future we will become more
and more reliant on CFD to predict the aerodynamic and propulsive forces and moments on
these vehicles, and accordingly, must accept the uncertainties in the predicted aerodynamic
and propulsion forces and moments as part of the risk of the flying the vehicle. One means
of reducing the inherent risk to the vehicle due to these uncertainties is to utilize an adaptive control law. However, it is imperative that the adaptive control law have theoretically
guaranteed stability and performance that can be validated prior to flight for the expected
variations in the aerodynamics and a set of defined actuator failures.
In a simulation study conducted by Wise,3 who simulated missile autopilots with and
without adaptive adaptive control and compared their robustness, it was concluded that
adaptive control has been shown to ...address a bigger class of uncertainties, as compared
to linear control. A separate study of the 1967 crash of X-15-3 conducted by Dydek,
Annaswamy and Lavretsky4 compared the performance and robustness of the Honeywell
MH-96 adaptive controller used on the actual vehicle to an adaptive controller designed
using the modern approach that utilizes Lyapunov stability theory. The crash and loss of
X-15-3 resulted when an 80% loss of control eectiveness in the right elevon was caused by
an electrical disturbance. The loss of control eectiveness in turn destabilized the vehicles
lateral-directional dynamicssomething that the Honeywell adaptive controller could not
accomplish. It was shown that the adaptive controller derived from Lyapunov stability
theory was robust to a wide range of uncertainties in the system and, in fact, was able to
maintain control and tracking performance under the same scenario as which caused the
vehicle to crash.
A.

Adaptive Control Architecture

The control law that will be implemented in this flight experiment is a direct model-reference
adaptive control law rather than an indirect adaptive control law. The dierences are that
a direct adaptive control law does not estimate the plant parameters in real-time. In a
direct adaptive law, the controller parameters are determined on-line by comparing the plant
response to that of a desired (ideal) reference model. In an indirect adaptive control law, the
plant parameters are estimated along with the computation of the controller parameters to
control the unknown plant. As a result, the indirect adaptive control law requires that the
estimation of the plant parameters converge to their unknown true values. This requirement
in turn necessitates a persistency of excitation condition on the reference input in order
to ensure convergence, and therefore is unreliable unless persistency of excitation can be
guaranteed.
For this vehicle, we employ a model reference adaptive control architecture that augments
a robust baseline controller, uBL with an adaptive increment, uad , such that the control input
to the vehicle u = uBL uad drives the tracking error to zero. In this case the tracking
error is defined to be the error between the plant and a known reference model. Augmenting
the nominal control input with an adaptive element provides robustness to uncertainties
that arise in the aero-propulsive system. This particular approach has been successfully
implemented and flown on several vehicles.5
7

To design the adaptive control law, the vehicle dynamics are expressed as a linear, timeinvariant system.
x = Ax + B(u + d(x))
(1)
where the state x(t) Rn , the control input u(t) Rm , > 0 Rmm is an unknown
high frequency gain, and d(x(t)) Rm is the matched uncertainty. It is assumed that
can be written as = Im + and that d(x) can be linearly parameterized as d(x) =
W (x)+(x), x Rn . A nominal/baseline control law for the system is designed assuming
that = Im and d(x) = 0 in Equation 1. The baseline feedback law is of the form

Downloaded by Alison Stewart on September 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-252

uBL = Kx + Kr r

(2)

The reference model is then formed by applying the baseline feedback law to Equation 1,
giving
x m = Am xm + Bm r
(3)
where Am = A BK and Bm = BKr .
The control law is written as u = uBL uadap , where uadap is the adaptive input that
augments the baseline control law. The adaptive control input is

uadap = (Im + ) uBL + W (x)


(4)
is an estimate of , and it can be shown that
where
= W Proj(W
, (x)eT P B)
W
= Proj( ,
B T P eu)

(5)
(6)

and
remain
where Proj is the projection operator that ensures the adaptive gains W
bounded,and is as defined in Reference [6]. The learning or update rates are w and and
are chosen by the control law engineer. Furthermore, P = P T is the solution to the algebraic
Lyapunov Equation
ATm P + P Am = Q
(7)

where Q > 0 is specified. The error between the plant and the reference model is, e = xxm .
Note that the above description does not include the control input maximum and minimum
deflection constraints, but these will be incorporated into the final control law design to
ensure the deflection limits are obeyed and that closed-loop stability is ensured if control
eectors saturate.
B.

Inner-loop Control Law

The inner-loop control system was designed to track a reference angle-of-attack profile and
therefore only includes the longitudinal dynamics of the aircraft. For the control design,
the state vector is defined as xp = [ q ]T , where is the perturbation angle-ofattack, q is the perturbed pitch rate, and is the perturbed pitch attitude. The control
input is e , which is the elevator deflection measured with respect to the trim elevator

deflection. We define the error between the angle-of-attack and the commandedangle-ofattack as e = cmd and append the
integral angle-of-attack tracking error, e dt so
the state vector becomes xp = [ q e dt]T .
The linearized, longitudinal equations of motion for the baseline controller design are of
the form
x p = Ap xp + Bp u + Gr
(8)
where

1 (g/Vt0 ) sin 0
0
0
1
0
0
0


0
Ze /Vt0
0


0
M e
0
, Bp =
, and G =
0
0
0
0
0
1

(9)

Note that for a hypersonic vehicle, under certain assumptions,7 Mq , Zq , and M can be
neglected, which is the case here.
We used linear quadratic regulator theory (LQR) to design a single baseline control law
and reference model for the vehicle. For the design, the vehicle was trimmed for rotational
equilibrium (Q = 0) and linearized about the flight condition at the beginning of the test
window. The closed-loop poles of the reference model are shown in Figure 5. Note the
pole-zero cancellations near the origin.
3

Imag [1/s]

Downloaded by Alison Stewart on September 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-252

Z /Vt0

M
Ap =
0
1

3
25

20

15

10

Real [1/s]
Figure 5. (s)/cmd (s) Closed-Loop Poles

The adaptive control law utilizes a set of radial basis functions to approximate the para
metric uncertainties d(x). The vector elements of are defined as i (x) = exp ( c,i )2 /(2 2 ),
9

where is measured in degrees, the centers of the radial basis functions that were selected
are c,i = {2, 0, 2, 4, 6, 8, 10, 12} deg, and the variance is 2 = 3 deg2 . The learning rates
are chosen as W = 100 and = 10. Note that this particular control law implementation
does not account for control eector limits, but these will ultimately be included in the final
design of the adaptive control law for the FLiRV.

Downloaded by Alison Stewart on September 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-252

V.

Simulation Results

Two dierent cases were run to evaluate the adaptive control laws performance and
robustness to uncertainty, but it should be noted that this is in no way an exhaustive or
complete study of the control laws performance and robustness. Additionally, actuator and
sensor dynamics are not modeled. In the first case considered, the the vehicle response is
for the nominal aerodynamic model used to design the inner-loop control adaptive control
system (i.e., d(x) = 0 and = Im ). A second trajectory was run where the aerodynamics
were perturbed from the nominal. First, the unknown high frequency gain was changed to
reduce the control eectiveness by 20%. After 3 sec of simulation time, we applied changes in
the lift, drag, and pitching moment coecients. We eectively shift the open-loop systems
short period poles by appending the following moment increment: Cm = 0.2 + 104 .
The perturbation to the lift coecient was CL = 0.05 to model a loss of lift, and
the drag coecient perturbation added a significant increase in drag, which was CD =
1.2CDW B (M, ).
A.

Nominal Case

The results of the adaptive control laws tracking performance for the nominal aerodynamics
are shown in Figures 6-10. In this case, an angle-of-attack of 3 deg is commanded at t =
0, and an angle-of-attack of 0 deg is commanded at t = 10. It is evident that in the
case of the nominal aerodynamic model that the adaptive control law can easily track this
angle-of-attack commandwhich is expected. The control eort as shown in Figure 7 is
reasonable, and does not require a significant change to track the angle-of-attack command.
The adaptive gains are small, as is expected since this simulation is applied to the nominal
vehicle aerodynamics used to design the linear control law that was augmented. Finally, note
that in Figure 10 that the pitch rate is nearly constant at zero once the transients subside.
B.

Perturbed Aerodynamics

The results for this case show that the controller is still is able to track the commanded
sequence, which is identical to the one presented above. Once the uncertainties are activated
at t = 3, there is a small transient in the states, but they are quickly damped. The control
time history at this event increases about 27 degrees to a large positive deflection (trailing
edge down) once the the step change in the aerodynamics is applied, with the deflection
increasing about 27 degrees At t = 10, when the the angle-of-attack is commanded to zero,
the angle-of-attack exhibits an oscillation that takes about 7 sec to damp out. This is also
reflected in the Q, , and states as shown in Figures 15. The control input (Figure 12)
follows a similar trend, but is significantly larger than before. This is attributed to the
reduced control eectiveness and the change in the pitching moment of the vehicle. Once
10

3.5
3
2.5

, deg

2
1.5

Plant
Ref. Model

0.5
0
0.5
0

10

15

20

Time, sec
Figure 6. Nominal Aerodynamics: Time History

0
1

uB L
2
3

u, deg

Downloaded by Alison Stewart on September 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-252

4
5
6
7
8
0

10

15

Time, sec
Figure 7. Nominal Aerodynamics: Control Time History

11

20

1.5

x 10

1
0.5
0

(t)

0.5
1
1.5

2.5
3
3.5
0

10

15

20

Time, sec
Time History
Figure 8. Nominal Aerodynamics:

x 10

0
2
4
6

x(t)
W

Downloaded by Alison Stewart on September 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-252

8
10
12
14
16
18
0

10

15

Time, sec
Time History
Figure 9. Nominal Aerodynamics: W

12

20

4
3
2
1
0
1

, deg
Downloaded by Alison Stewart on September 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-252

Q, deg/s

, deg

, deg

5
0

10

15

20

Time, sec
Figure 10. Nominal Aerodynamics: Vehicle Attitude Time History

the transients subside, the vehicle tracks the reference model rather well. Additionally,
and W
remain
despite the fact that the vehicle aerodynamics are perturbed, the gains
gains are approximately twice as large as in the
small (Figures 13-14. However, the W
nominal case and are much more dynamic as compared to the nominal case, which is not
is two orders of magnitude larger, although it is still small, and
unexpected. The gain
appears to be constant at t = 20 sec.

VI.

Summary

The Hypersonic International Flight Research Program (HIFiRE) is an international


collaboration between the U.S. Air Force Research Laboratory and the Defence Science
Technology Organisation in Australia. The HIFiRE program is investigating basic hypersonic
phenomena and experimental technologies critical to the development of next generation
aerospace vehicles and the understanding of the environments in which they must operate.
The paper describes the mission and trajectory of HIFiRE Flight 6, which is a flight research
experiment designed to evaluate the performance of an adaptive control law on a class of
hypersonic vehicles. For this flight, a Terrier-Oriole sounding rocket will be used to insert the
flight research vehicle to its test window. The flight research vehicle will fly a pre-determined
trajectory where aerodynamic uncertainty is expected to be significant. An initial design of
a model reference adaptive control law has been performed to assess angle-of-attack tracking
13

4
3.5
3

, deg

2.5
2
1.5
1

0
0.5
1
0

10

15

20

Time, sec
Figure 11. O-Nominal Aerodynamics: Time History

25
20

uB L
15
10

u, deg

Downloaded by Alison Stewart on September 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-252

0.5

Plant
Ref. Model

5
0
5
10
15
0

10

15

Time, sec
Figure 12. O-Nominal Aerodynamics: Control Time History

14

20

2.5

x 10

(t)

1.5

0.5

0
0

10

15

20

Time, sec
Time History
Figure 13. O-Nominal Aerodynamics:

0.045
0.04
0.035
0.03
0.025

x(t)
W

Downloaded by Alison Stewart on September 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-252

0.02
0.015
0.01
0.005
0
0.005
0

10

15

Time, sec
Time History
Figure 14. O-Nominal Aerodynamics: W

15

20

Downloaded by Alison Stewart on September 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-252

, deg
Q, deg/s

, deg
, deg

6
0

10

15

20

Time, sec
Figure 15. O-Nominal Aerodynamics: Vehicle Attitude Time History

performance, via simulation, for the vehicles longitudinal dynamics. In particular, the
control law was tested in simulation on both the nominal aircraft aerodynamics, and with
reduced control eectiveness and perturbed aerodynamics that represent a significant change
in the aerodynamics of the system, and in fact represents a significant instability in the shortperiod dynamics.

VII.

Acknowledgement

Approved for Public Release; Distribution Unlimited. Case Number 88ABW-2011-6376.

References
1

Smith, T. R., Bowcutt, K. G., and et.al., HIFiRE 4: A Low-Cost Aerodynamics, Stability, and Control
Hypersonic Flight Experiment, AIAA-2011-2275 , 2011.
2
Baumann, E., Bahm, C., Martin, J., Bose, D., Beck, R., and Strovers, B., The X-43A Six Degree of
Freedom Monte Carlo Analysis, AIAA-2008-203 , 2008.
3
Wise, K., Robust Stability Analysis of Adaptive Missile Autopilots, AIAA-2008-6999 , 2008.
4
Dydek, Z., Annaswamy, A., and Lavretsky, E., Adaptive Control and the NASA X-15-3 Revisited,
IEEE Control Systems Magazine, Vol. 30, No. 3, June 2010, pp. 3248.
5
Wise, K., Lavretsky, E., and Hovakimyan, N., Adaptive Control of Flight: Theory, Applications, and
Open Problems, Proceedings of the 2006 American Control Conference, 2006.
6
Pomet, J. and Praly, L., Adaptive Nonlinear Regulation: Estimation from the Lyapunov Equation,

16

Downloaded by Alison Stewart on September 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2012-252

IEEE Transactions on Automatic Control , Vol. 37, No. 6, June 1992, pp. 729740.
7
Rudd, L., Hodgkinson, J., Parker, R., and Tarpley, C., Hypersonic Stability Derivative Modeling
Issues, AIAA-2010-7929 , 2010.

17

S-ar putea să vă placă și