Chinese Journal of Aeronautics, 2013,26(4): 909917

Chinese Society of Aeronautics and Astronautics

& Beihang University

Chinese Journal of Aeronautics

cja@buaa.edu.cn
www.sciencedirect.com

Gain self-scheduled H1 control for morphing aircraft in

the wing transition process based on an LPV model
Yue Ting

a, *

, Wang Lixin , Ai Junqiang

School of Aeronautic Science and Engineering, Beihang University, Beijing 100191,

b
China The First Aircraft Design Institute of AVIC, Xian 710089, China
Received 21 May 2012; revised 7 September 2012; accepted 10 January 2013
Available online 2 July 2013

KEYWORDS
Gain self-scheduled; H1 robust control;
Linear parameter varying; Morphing aircraft;

Wing transition

Abstract
This
article
investigat
es
gain
selfscheduled
H1 robust
control
system
design for
a tailless
fold-ingwing
morphing
aircraft in
the wing
shape
varying
process.
During
the wing
morphing
phase,
the
aircrafts
dynamic
response
will
be

co
mb
at
1. Introduction
mis
sion
Morphing aircraft can automatically s by
change its aerodynamic configuration to usin
adapt to different flight environments and g

advanced

governed by time-varying aerodynamic forces and

moments. Nonlinear dynamic equations of the
morphing aircraft are linearized by using Jacobian
linearization approach, and a linear parameter
varying (LPV) model of the morphing aircraft in
wing folding is obtained. A multi-loop controller for
the morphing aircraft is formulated to guarantee
stability for the wing shape transition pro-cess.
The proposed controller uses a set of inner-loop
gains to provide stability using classical
techniques, whereas a gain self-scheduled H 1
outer-loop controller is devised to guarantee a
specific level of robust stability and performance
for the time-varying dynamics. The closed-loop
simulations show that speed and altitude vary
slightly during the whole wing folding process, and
they converge rapidly after the pro-cess ends.
This proves that the gain self-scheduled H 1 robust
controller can guarantee a satisfactory dynamic
performance for the morphing aircraft during the
whole wing shape transition process. Finally, the
flight control systems robustness for the wing
folding process is verified according to
uncertainties of the aerodynamic parameters in
the nonlinear model.

and E-mail addresses:

yueting_buaa@sina
actuators.
.com
(T.
Yue),
bhu_wlx@
tom.com (L. Wang).
Peer review under
*
responsibility of
Corresponding author.
Editorial Committee
Tel.: +86 10 82338821.
of CJA.

2013 Production and

hosting by Elsevier
Ltd. on behalf of CSAA
& BUAA.
Open access under CC BYNC-ND license.

materials

1,2

Pro
duct
ion

and hosting by Elsevier

Co
mp
are
d
to
con
ven
tion
al
fixe
dwin
g
airc
1000-9361 2013 Production

a
n

raft, morphing air-craft characteristics.

possesses
multi- The
wing
objective adaptability and transition process
higher
combat is
obviously
3,4
effectiveness.
The complicated and
typical
objective
of very important for
morphing aircraft de-sign morphing aircraft.
includes enhancing flight Generally, the wing
performance and combat
effec-tiveness, but not morphing
will
improving
the
flying approach
quality. So application of involve large rigidmorphing
techniques body motions of
may bring disadvantage the wing structure.
dynamic
to the air-crafts dynamic The

d hosting by Elsevier CSAA &

Ltd. on behalf of
BUAA. Open

access under CC
BY-NC-ND license.

response
of
morphing aircraft
will be governed
by
time-varying
aerodynamic
forces
and
moments
which
are related to the
wing
shape.
Dynamic models
for
morphing
aircraft must take
into account the
dynamic

http://dx.doi.org/10.1016/j.cja.2
013.06.004

910

T. Yue et al.
exploits

all

h varies in a polytope H, that

is h e H = Co{h1, h2, . . ., hr};
the plant is further assumed to
be polytopic, i.e.,
12
3
CAh B1h
D B2h
D

available

coupling between wing area and mass distribution change, aerody- information on h(t) to adjust
namics, structure, and control forces. Thus, the wing morphing the current plants dynamics.

This provides smooth and

57

process is a complicated time-varying dynamic system.

As wingautomatic
gain
selfconfiguration parameters and aerodynamics do vary in large ranges, scheduled with respect to
1 h
the dynamic response in wing morphing and the final bal-anced the varying parameter h(t).
6 C2 h
kinetic parameters will also exhibit large variations. Further-more, the
We consider LPV plants
4

aircraft needs a long time to achieve a new state of level flight without
P(h) with state-space
A
control. Consequently, the flying quality of morp-hing aircraft may
equations
Co 82 C1ii
deteriorate and flight safety may be threatened. To guarantee
C
>
2i
satisfactory flying quality, a morphing aircraft capa-ble of shape 8 x Ahx B1hw B2hu
>
11
12
reconfiguration requires a flight control system to maintain stability > C D D
:
< y 2 h x 21 h w 22
during the morphing transition phase.
<6

C1

>

1i

12

22

11i

21i

2i

12i

22i

7 2
5
h

3
7

; i 1; 2;

where Ai, B1i, . . ., denote

the values of A(h), B1(h), . . .,

n
at the vertices h = hi of the
x e R , disturbance
mainly focuses on static configuration control.
There is a with state
parameter polytope.
m1
m2
w
2
R
,
control
input
u
2
R
notable lack of published work on flight control design for
p1
morphing aircrafts wing transition phase. When the aerody-namic , performance output z 2 R ,
p2
shape is changing, the controller should adapt online to maintain and measured output y 2 R .
The
plants
P(h)
can
be
stability in shape transition between different con-figurations. described as
Since the wing transition process is time-varying, it is obviously
P P
quite difficult to use general control techniques to control such a
11
12
2 Ah B1
P
complex dynamic behavior. One possible ap-proach is to assume
P P21
22
C1h D
that the time-varying dynamics could be represented by a linear
parameter varying (LPV) plant model that approximately captures
4
We seek an LPV controller K(
the wing transition phases com-plex behavior. In this LPV
framework, nonlinear dynamic equations of morphing aircraft xK
AKhxK BKhy
C h xK
DK
h y
could be simplified and trans811

formed to an LPV model. Then gain self-scheduled control

1214
technique
based on the LPV model could be used for wing
that

C2

21

guarantees
H1
transition control. The gain self-scheduled control technique performance for the closedbased on the LPV model can rapidly change controller param- loop system in Fig. 1. The
eters to adapt to the aircrafts dynamic response. Additionally, this transfer function from w to z
control approach can guarantee stability of the closed-loop of the closed-loop system is

system. Therefore, the gain self-scheduled control tech-nique can

be utilized to solve flight control design problems for the wing Tzws P11 P12KI
1
P22K P21
5
transition process of morphing aircraft.
This article focuses on gain self-scheduled H 1 robust control
design during the wing folding process of a tailless folding-wing Definition 1. 15A matrix
morphing aircraft. According to the properties of wing folding polytope is defined as convex
process, longitudinal nonlinear dynamic equations of the morp- hull of a finite number of
matrices N with the same
hing aircraft in wing shape varying are simplified and trans-formed dimensions,i i.e.,
to an LPV model. Then, a multi-loop controller designed based on
(
; rg :
the LPV model is presented. It is found that this control approach CofNi; i 1; 2;
can successfully maintain stability for the morphing aircraft in the
a
i
whole wing transition process.
r1 i

2. Gain self-scheduled H robust control for LPV system

X
where ai is the weighting ratio.
For the LPV system Eq. (2), if:

the
parameter
The class of finite dimensional linear systems whose state-space (A)
dependence is affine, that is,
matrixes depend continuously on a time varying parameter vector the state-space matrices A(h),
h(t) is called linear parameter varying. In state-space form, an B1(h), B2(h), C1(h), C2(h),
LPV system model can be expressed as
D11(h), D12(h), D21(h) and
D22(h) depend affinely on h;
x Ahtx Bhtu
(B) the tine-varying parameter
y Chtx Dhtu

description
is
where the state matrixes A, B, C, Dan
explicit
vary with h(t), u is the con-trol input.function of h(t).
It is a time-varying vector that can Here
we
consist
of
system
outputs,focus on the
exogenous inputs, or combinationsstandard
H1
of both. LPV systems are linear control problem
systems where the state-spaceof LPV system.

As shown in
Fig.
1,
the
resulting
LPV
controller K(h)

F
i

;r 9
>
=

>

Current research on flight control design for morphing air-craft

11

D21

When the
wing based
shape on
is an LPV
handmodel
side
Gain self-scheduled H1 control for morphing aircraft in the wing transition
process
911
the
dynamic
In the sequel the standard H 1 control problem can be de-changing,
of Eq. (11) can be
fined as: finding an internally stabilizing controller K that response of the morphing
expressed as
makes the closed-loop H1 gain from w to z less than c. Ifaircraft will be dependent on
8
aero-dynamic
Tzw(s) denotes the closed-loop transfer function w to z, thistime-varying
>
< Fx D cos a L sin a T
control objective can be formalized as
forces and moments, which
are both functions of the wing
16
shape. In Ref.
the folding15
wing
morphing
aircraft is reIt is known from Ref.
that there exists an LPV controller
Eq. (4) guaranteeing quadratic H1 performance c along allgarded as a variable geometry
parameter trajectories, if and only if there exist two symmetricrigid body, and a six-DOF
matrices R e Rnn and S e Rnn satisfying the system of 2r + 1 non-linear dynamic model in
linear matrix inequalities (LMI).
the wing folding process is
If h is in the polytope Co{h1, h2, . . ., hr}, that is,
founded. It also shows that
( r
)
the aerodynamic forces and
r
h 2 H : i 1 aihi : ai P 0; i 1 ai 1
9 moments of the folding-wing
morphing
aircraft
almost
X
X
linearly vary with the wing fold
Then the LPV controllers state-space matrices are given by
angle hfold in wing folding.
A B
r
Ki
Ki
AK h t
BK h t
Hence, we can regard the

D
a t
:
CKht DK h t
Ki
10 morphing process as an LPV
i 1 i CKi
system. In this LPV frame
X
work, the nonlinear dynamic
where AKi, BKi, CKi, DKi can be obtained off-line, and AK(h(t)),equations of the morphing airBK(h(t)), CK(h(t)), DK(h(t)) will update dependently on the
craft will be simplified and
parameter h(t) in real time.
transformed to an LPV model
LPV
modeling
3. Longitudinal LPV model in wing shape varying for folding- by
1720
approaches.
wing morphing aircraft

kTzwsk1 < c

mg sin h
> Fz D sin a L cos a mg cos h

My MA gSz sin h
where D, L, T are the drag,
lift and thrust, MA is the
pitching moment caused by
aerodynamic forces, a the
angle of attack, h the pitch
angle, Sz the z-component
of static momentum S = * r
dm in fixed-body axes, and
g the acceleration due to
gravity, respectively. The lift,
drag, and pitching moments
in Eq. (12) are given as

Here we only consider

the longitudinal motion of
In this article a folding-wing morphing aircraft which has a tail-less
the air-craft during the wing
flying wing configuration is studied (shown in Fig. 2). The wingsfolding
process.
The
nonlin-ear
include inner wings and outer wings, which can be folded by longitudinal
smart actuators to change their shape. In the wing transition equations of motion in the
wing folding process can be
process, the inner wings rotate and the outer wings keep level.
16
expressed as
Iy

mqVz

Sz 2
Sz

>

>

z z

<
>

>

>
>
>
>
>
>

SC

MA

qV2 ScC m

>
>

>

where
is the air density, S the wing ar
>
q
:

aerodynamic chord. The lift,

drag and pitching moment
coef-ficients are given as

CL CLa; V CL0 CLaa CLVDV=V

CD

>

<

Cm

CD a; V

D0

CDaa CDa2 a2

q Sz

Sz

Fz

_
2qSz

Fx

I yq

>

>h
>
_

mIy

Sz

combining Eqs. (11)(14) and

using Jacobian linearization
approach, we get the LPV
model presented in statespace form as

SzqV
z

Fig. 2 Folding-wing morphing aircraft.

The variables in Eq. (11) are introduced in Ref. It is as-

sumed that the thrust line

passes through the origin of
the fixed-body axes. The
forces and moments in right-

Cm a; q; V; de Cm0 Cm a C D
mV

>

>
:

12

The variables in Eq. (14)

16
are given in Ref.
By

mqVz

>

>

>

<
>

>
>
>

>
>

:
2

mIy Sz2

>D

>

>

>
>

>
>
>
>
>

8
S qV

Iyq

S
>
mIy
>
>
>
>
1
> Vz
mqVx
>

> m
>
>

2qSz Fx

>
>

>

L 1 qV2SCL

8 Vx mIy
>
>

912

T. Yue et al.

u
2w 3
6q 7
6_ 7
6

u u0

Dde
2 w w0 3
6
Aht
7 Bht DdT Wht 15
q
6h
h0 7
6

4 5
4
5
where u, w are the x and z components of airspeed in fixed-body
axes, u , w the x and z components of balanced airspeed in

0 0

fixed-body axes before wing folding, V u

w , DdT is the

propulsion control, and h0 the initial pitch angle. W(h(t)) is the force
and moment variations affected by wing folding, which can be
considered as a disturbance source in the wing morphing dynamic
response. In Eq. (15) the state-space matrices vary with

hfold, and terms in the matrices are given in Appendix A. When

using Jacobian linearization approach to transform
nonlinear model to an LPV model, the dynamic response of
morphing transition phase produced by the two models should be
similar, i.e., the obtained LPV model should be rational.
Parameter comparisons of the dynamic response of the two
different models are shown in Fig. 3, in which it can be seen

Fig. 3 LPV vs nonlinear model

simulations
of
morphing
aircraft.

that distinctions between the

dynamic responses in LPV
model and nonlinear model
are small. The LPV model is
able to cap-ture the dynamic
behavior and match the
nonlinear model. Therefore,
the control system design of
the morphing aircraft in wing
folding can be based on the
LPV model. Keeping the
altitude and speed constant
during wing shape varying can
be the control objective.
According to the property of
the LPV model, LPV control
techniques could be chosen to
guarantee smooth transition
between
different
configurations.
The
initial
dynamic
response of the wing folding
process mainly represents
variations in the aircrafts
pitching
moment.
The
aerodynamic forces acting on
the aircraft change a little.
Then the speed and altitude
vary only slightly, and a, h, q
all increase as the pitching
moment in zero lift increases
when the wing begins to fold.
When the wing continues to
fold, a and q gradually
converge,
the
altitude
increases and the speed
decreases, indicating that the
aircraft begins to climb with
decreasing
speed.
Every
parameter will converge to a
stable va-lue after a long
period of time and the aircraft
will enter a new state of
balanced flight.
In the wing folding process
of the tailless folding-wing aircraft, the altitude, speed and
pitch angle vary largely, and
the aircraft will need a long
time to achieve new stable
flight. In order to guarantee
satisfactory flying quality and
safety in the process of wing
folding, a flight control system
is required.

4. Multi-loop controller
design and simulation
for the wing morphing
process
We know from the dynamic
response of the folding-wing
morp-hing aircraft in transition

phase that the altitude and speed do change a lot. Moreover, it takes keep the altitude and speed
a long time for the aircraft to achieve new balance in stable flight. To stable when the system x
ensure that the morphing aircraft can maintain stable flight during Ahtx Bhtu is
wing shape varying, the wing morphing process needs a flight control disturbed by w(h(t)).
system to obtain a prospective objective. In this article, the control
According to the property
objective is to keep the altitude and speed constant during the of the LPV system, the inner
transition from extended-wing configuration to folded-wing loop of the control system is a
linear
quadratic
output
configuration.

which can be solved by using

a convex hull algo-rithm. The
multi-loop controller is shown

tracking
performance
x Ahtx Bhtu B1wht
16 characteristics for the linear
system x Ahtx B
4
2
41
44
42
4
where x e R , u e R , w e R , A e R , B e R , B1 = I . Considering
htu. The outer loop is
w(h(t)) as a disturbance, then the system can be
composed of a gain selfregarded as an LPV system x Ahtx Bhtu with a
scheduled robust H1 controller

The controller design

objective in wing transition
phase for the folding-wing
morphing aircraft is to keep
the altitude and

The longitudinal LPV model of the morphing aircraft in feedback con-troller which
provides stability, as well as
morphing can be rewritten as

dis-turbance w(h(t)). Therefore, the flight control objective is to

in Fig. 4.

4.1. Inner-loop linear

quadratic optimal control
with output feedback

0
Gain self-scheduled H1 control for morphing aircraft in the wing transition process based on an LPV model
2 0
6 0
6

Fig. 5 Inner-loop output

feedback control
structure.

0
0 3
0 7

6 0
6

0
0

2
0

optimal controller with output feedback, as shown in Fig. 5.

The augmented state equations of the system shown in

Fig. 5 can be expressed in the form of
x
8

Ax

>

Bu

Cx

Fr

Gr

< z Hx

>

>z

>
<
>

&

18

DH Du

d T &T

u de

6
6

a3 1

6
6
6
6

1 0

u0

0
0
0
0

13

14

32
33 34
a
a
a
22

23

42

a41

6
6

12

24

43

44

11

12

31
32
b
b
21

22

41

42

0
20

6
6
6

0 07
7
0 0
7

0 07
0 0 0 77

0
1

7
0 07
0

7
7

1 0 0

0 0

0 1

0h

eH

0e

In the augmented description, the gai

Keu and Keu are unknown, and need to
acceptable closed-loop
performance.

0 0
0 03
0

0 0 0 1

Kq K

With this structure, the plant matrices are given by

11

G
0 0 0 0 0 0 0 0 0
According to Fig. 5, the control input
u KLQy

a21

0 0 0

1 0 0 0 0 0 0 0 0
0 0 1 0 0 0 T

&

>

0 0 0

>

F 0 0 0 0 1 0
0 0 0 0 0 0

>
where

Du Dw Dq Dh DH de

8 y Dq Dh eH eH
eu eu & T

17

0 1 0 0 0 0 0 0
0 0 1 0 0 0 0 0 3
0 0 0 1 0 0 0 0 7
7
0 0
0
0
0
0 1 0

6
6
6

7
7
7
7
7
7
7

0 5

0
2 0
C
6 0
6 0

speed constant. The inner-loop of the control system is an LQ

20

4 0
Fig. 4 The multi-loop control structure.

0 7

6
6
6
6
6
6
6

913

19

To solve the output gain

21
problem, Stevens and Lewis
de-vised the conversion from
tracking
to
a
regulator
problem, using a new set of
deviation variables. To make
both the er-ror deviation e~t
and the steady-state error e
small, the out-put gain K is
selected to minimize the
performance index (PI)
1

T
T
T
J 2 Z0 e~ e~ u~ Ru~dt 2 e Ve
with R > 0, V P 0. The
optimization problem can be
solved
numerically
using
an
iterative procedure through
Lyapunov equations.22,23

As the state-space
matrices of the LPV model
change with

the wing fold angle hfold, the

optimal output gain KLQ will also
change with corresponding hfold.
Nevertheless, a time-invariant

gain KLQ is preferred, although only a suboptimal solution would be

achieved in such a case. The performance of the closed-loop system
in the wing folding process can be achieved by outer-loop selfscheduled H1 robust control design. Here we select the optimal
output feedback gain in extended-wing

configuration as the inner-loop gain:

KLQ

16 5
0

68: 7 0 14 0 03
0

:0

:0

15:8

18:9
27

914

T. Yue et al.

4.2. Outer-loop gain self-scheduled H1 robust control

following formulas for the state-space data of the LPV controller are obtained:

The closed-loop system with LQ output feedback inner-loop

controller is found to be
x Ac h t x B h t u B1w h t

8 z Hx
y Cx

where

>

32

CKi DKi

X
It is easy to check that ai are convex coordinates satisfying

h 2 H : f

ah :a P

0;

in

1g. The resulting poly-

i1 i

performance

a32max & 0:0618

4.3. Closed-loop simulation

The resulting gain self-scheduled H1 robust Controller based
on the LPV model is applied at the morphing aircrafts wing

22

tor of the LPV plant is denoted as ht : a32 . And

a
a
a
22
22min
22max
1:3392 0:5212

30

0:0344 &

&
&
So h(t) ranges in the polytope H e Co{hi, i = 1, 2, 3, 4}. The
vertices hi are the values of h(t) at the four vertices of the
h1
22min

Ki

eter variations.

a t

i 1 i

u KLPVhy
29
to make the altitude and speed remain constant during the
wing
morphing
process.
The reference
input
T
r rDH rDu & 0, and the outer-loop LPV control structure
is shown in Fig. 6.
When using convex hull algorithm to design the LPV controller, if there are n variable elements in Ac(h(t)) and B(h(t)),
n
the convex hull will have 2 vertices. The calculation will be
complicated if n is big. Therefore, we assume that the elements
in Ac(h(t)) and B(h(t)) do not change as they vary in very small
range. In other words, we only consider the elements that vary
in large ranges during the wing folding process.
From the simplified LPV equations of morphing aircraft we
find that the variable elements a22 and a32 in Ac(h(t)) vary in
large ranges, which both depend on the aerodynamic changes
during the wing folding process. Hereafter, the parameter vec-

parameter

Ki

over the entire parameter polytope H and for arbitrary param-

The gain self-scheduled H1 control problem consists

finding an LPV controller

a32min

topic LPV controller enforces stability and H

Acht Aht BhtKLQC:

:
DK h t

i1 i i

a32

CKht

28

>
<

BK h t

AK h t

box:

22min

a32min

22max

a32min

h3

22max

; h4 a32max
.It is clear that the state-space maa32max
trix Ac(h(t)) ranges in a polytope of matrices whose vertices
are the images of the vertices h1, h2, h3, h4. In other words,
Acht 2 CofAi : Ahi; i 1; 2; 3; 4g

31

Finally, a formal expression for the LPV controller is

derived by solving the convex decomposition problem. The

Fig. 6 Outerloop LPV control

structure.

Fig. 7 Closed-loop response in morphing process.

Gain self-scheduled H1 control for morphing aircraft in the wing transition process based on an LPV model
wing folding process.

shape transition phase. The response of the aircraft in closedloop simulation is shown in Fig. 7. It is clear that the speed is
almost constant and the altitude decreases about 0.7 m to the
maximum during the wing folding process. They can both converge in 10 s after the wing finishes folding. Since the wing
area decreases after it folds, the angle of attack will increase
to achieve a new balance. In addition, the changes in elevator
deflection and throttle are both within acceptable ranges. It is
obvious that the gain self-scheduled H1 robust controller
based on the morphing aircrafts LPV model can eliminate the
disturbance caused by wing folding and guarantee con-stantspeed and altitude flight during the wing transition process.

915

4.4. Robustness verification

In
this
article
the
aerodynamics in wing folding
obtained by CFD is assumed
to be quasi-steady, and
unsteady aerodynam-ics is
ignored. Consequently, the
aerodynamics in the nonlinear model of the morphing
aircraft is not modeled
precisely, and not exactly the
same with aerodynamics in
actual flight conditions. In
other words, the aerodynamic
model of the wing folding
process is uncertain within a
range. Thus it is impor-tant for
the controller to have the
ability to provide stability in
spite of modeling errors due
to unmodeled dynamics and
plant parameter variations.
The unpredictable manner
of the model can be described
by parameter perturbation.
Here we use the parameter
perturba-tion
of
the
aerodynamic
force
and
moment coefficients to describe
model
parameter
perturbation.
The
aerodynamic
coefficient
perturbation is dependent on
the angle of attack. When the
angle of attack is bigger, the
perturbation will be greater.
Thus the perturbations of
aerodynamic force and moment coefficients are set to be

p
0:8 0:5180 a CL
8

p
> CD0 1:2 0:5 180 a CD
>
>
< Cm0 0:8 0:5 p aCm
CL

>

>

180

>
>

>
:

where C are the perturbation

values, and C the rated
values. According to the
aerodynamic parameter
uncertainties in the model, the
closed-loop response of the
morphing aircraft during the
wing folding process is shown

Fig. 8 Closed-loop response in parameter perturbation during

in Fig. 8. It is seen that the

parameter amplitude of the
closed-loop response increases slightly due to the
aerodynamic parameter

perturba-tion. The motion of the morphing aircraft converges

slower after wing folding is completed. Even so, the flight control
sys-tem can still ensure that the speed variation in the wing morphing process is small, and that the altitude varies less than 1 m.
This means the gain self-scheduled H1 robust control system can
maintain good robustness under modeling uncertainty of

aerodynamics.

5. Conclusion
(1)The

LPV model for the folding-wing morphing aircraft in

wing folding is derived from longitudinal nonlinear dynamic
equations by using Jacobian linearization approach. The

LPV model includes the

variants
of
aerodynamic force and
moment caused by
wing folding, which can
be regarded as a
disturbance
source.
The dynamic response
is simulated using the
nonlinear model and
LPV model. It is shown
that the longitudinal
LPV model obtained by
Jacobian linearization
approach can capture
the morphing aircrafts

complex behavior in the

wing transition process.

(2)In order to ensure that

the morphing aircraft

flies at a given altitude
and speed in the wing
folding process,
an inner-loop optimal
quadratic
output
feedback and outerloop gain scheduled H1
robust controller based
on convex optimization
algorithm is designed.
The sim-ulations show
that
the
control
objective of maintaining

916

T. Yue et al.

speed and altitude change only slightly when the wing

folding process can be achieved via the multi-loop con-

Sz
a31 mIy

>

Sz2

>

trol approach. The morphing aircraft can rapidly return

>

>

>

>

>

qV 0SCDV

Da 0

Sz

mIy

>

to stable flight when the wing completes folding. In addi-

>
>
>

>

tion, the multi-loop flight control systems robustness is

verified according to the aerodynamic parameter uncer-

qu0Sc Cm0

>

ma

qV0cSCmV

>

>
>
>
>

C a

qu0S CD0

C C a
L0 ma 0

La

mde e0

>

>
>
>

z
>

tainties in the wing morphing dynamic model.

>

a32

>

>
>
>
>
>

mI

>
>

qV0SCDa
<

> 2

D0

qw0S

C a

Da 0 2

mIy Sz

>
>

Acknowledgements

C ma

>

>
>
>
>

qV0cSCma

qw0Sc Cm0

L0

ma 0

mde e0

La

>
>
>

Sz

>

This study was co-supported by China Postdoctoral Science

>
>

mI

>

>

mw0

a3 3

>

2Sz

Iy

mI

z2

Sz w0

2
qV0c SCmq

>

Foundation (Nos. 20110490259, 2012T50038). >

>
>
>
>

Sz

>

2 mg cos h0

34

gS cos h

>

>

mIy S

>
>

mIy

>

>

Appendix A

>
>

A4

:
a

11

8
>

12

a21

>

>
>
>
>

>

>
>
>
>
>
>
>
>

23

42

41

11

b 11

21

24

Sz
2

mIy

43

44 7

12

qV cSCm
0

de

11

>
>

>
>

Lde

m2

Sz

>

22 3

qV SC

b21

>
>
>

>

A5

de

mIy

2
1 qV SCD

Iy
2

5>

43

a 34

33

42

1
;

6a
6

24

0;

41

14

>

>

22

a 32 a

31

>

13

0
;

>

A1

>B

>
>

b31

>

31

32

>

>
>

>

de

SCD

S 2

mIy

>
>
>

cSCm

de

Sz2

mIy

7>
b

7
7

41

42

>b

>

>
5>

Iy

<

w
11

>

dp

12

>

>

>

mIy

>
>

>

<
>
6
6
>

>

>W
>
>

Sz

>

21

31

>b

>

22

>
>

>

>

>
>
>
>
>
>

S
z

>

>
>

41

>

>
>
>
>

>

b32

dp

>

>
>
>
:

mI

>
Iy

>

>

b42

>

11

8
>

mIy

A6

Sz

qV0SCDV

>

C a

qu0S CD0

Iy

Da 0

>

11

mI

>

>

>

>
>
>

qV0cSCmV

>

>
>

S
z

mIy

qV

S CD0

T0

Da 0

mg sin h0

ma

>

qu0Sc Cm0

L0

ma 0

>
Sz

mde e0
>

CL

>

>

a
>

>
>

>

mI

>

>

Iy

S z

=m1

1x

2S S

1z

2x

2
qV cS

=m2

2S S

2z

>

>

>

>

>
>

Sz

>

>
>
>

>

a12

>
>

>
>

mIy

>

>

>
>
>
>

>

>

>

>

>
>
Sz >

qV0SCDa

>
>

>

>

Sz

qw0S CD0

Da 0

2
mIy
>

<

L0

ma 0

>

S CL0

CL

>

w Sc C
ma

m0

CL

mg cos h0

a0

>

cSC

q V0

gSz sin h0

e0

2
qV

w21

>

>

Sz

>
>
>

C a

mde

>
>

>

ma

CLa

>

>

>

m0

>

C a

>

L0

>

C mde e0

ma 0

<

Sz

>
w

SC

D0

T0

3
1

Da 0

mIy

Sz

mg sin

>

Iy

>

>

>

>
>

>

mw0

a 13

>

2Sz

>

>

>

mI

>

>

>

1>

>
>

>

>

>
>

mIy

>

w
z

2
V

>

2x

2z

S
z

mIy

c SC

>
mq

>

>

qV cS
0

ma

m0

>

L0

>

>

C La
a

>

>

>

gS

e0

>

2S S
1z

>

1x

=m2

2S S

>

>

>
>

=m1

>

sin h0

>

>
>
>

>
>

mg cos

a14
2

>
gS cos

0
2

>
>

>

>

>

S z mIy

mIy

>

Sz>

>

>
>

A7

>

>

>

>

>

A2

w41

1 1

>

>

>

>
<

Silcox RJ, et al. The aircraft morphing, program. AIAA-1998-

1927; 1998.

u0

23

2. Wilson JR. Morphing UAVs change the shape of warfare. Aerosp

g sin

>a

>
>
>
>
>

A3

m 2

1. Wlezien RW, Horner GC, McGowan AR, Padula SL, Scott MA,

C Daa0

w S C D0

References

q V SC La

22

>

m 2 qV0SCLV qu0SCD0 CDaa0

a21

>a
>

24

Am 2004;42(2):289.
h0

>

:
Gain self-scheduled H1 control for morphing aircraft in the wing transition process based on an LPV model
2008;31(6):168799.

3. Bowman

JC, Plumley RW, Dubois JA, Wright DM. Mission 12.

Spillman MS. Robust
effectiveness comparisons of morphing and non-morphing vehilongitudinal flight control
cles. AIAA-2006-7771; 2006.
design
using
linear
varying
4. Yue T, Wang LX, Ai JQ. Key technologies in morphing aircraft parameter
design. Flight Dyn 2009;27(5):610 [Chinese].
feedback.
J
Guidance
Control
Dyn
5. Seigler TM, Neal DA. Analysis of transition stability for morp- 2000;23(1):1018.
hing aircraft. J Guidance Control Dyn 2009;32(6):194753.
Biannic J, Apkarian
6. Obradovic B, Subbarao K. Modeling of flight dynamics of 13.P, Garrard
WL. Parameter
morphing-wing aircraft. J Aircraft 2011;48(2):391401.

7.

Obradovic B, Subbarao K. Modeling of dynamic loading of

morphing-wing aircraft. J Aircraft 2011;48(2):42435.

8.

Seigler TM, Neal DA, Bae JS, Inman DJ. Modeling and flight control

of

large-scale morphing aircraft. J Aircraft 2007;44(4):107787.

9.

Ataei-Esfahani A, Wang Q. Robust failure compensation for a

morphing aircraft model using a probabilistic approach. IEEE
Trans Control Syst Technol 2007;15(2):32431.

10.

Valasek J, Tandale MD, Rong J. A reinforcement learningadaptive control architecture for morphing. AIAA-2004-6220; 2004.

varying control of a highperformance

aircraft.
J
Guidance
Control
Dyn
1997;20(2):22531.

14.

Natesan K, Gu D,
Postlethwaite I. Design of
static H1 linear parameter
varying
controllers
for
unmanned
aircraft.
J
Guidance
Control
2007;30(6):18227.

Dyn

11. Baldelli DH, Lee D, Sanchez Pena RS. Modeling and control of 15.
an

aeroelastic morphing vehicle. J Guidance Control Dyn

Apkarian P, Gahinet
P, Becker G. Self-scheduled
H1

control of

linear

917

parameter varying systems:

a
design
example.
Automatica
1995;31(9):125161.

16.

Yue T, Wang LX, Ai JQ. Longitudinal multibody dynamic 22.

Nelder JA. A simplex
method
for
function
characristics of Z-wing morphing aircraft. Acta Aeronaut Astrominimization. Comput
J
naut Sin 2010;31(4):67986 [Chinese].
1964;7(4):30813.

17.

Marcos A, Balas G. Linear parameter varying modeling of the

Boeing 747100/200 longitudinal motion. AIAA-2001-4347; 2001.

18.

Kumar A, Andersen MR. A comparision of LPV modelling

techniques for aircraft, control. AIAA-2000-4458; 2000.

19.

Marcos A, Balas GJ. Development of linear-parameter-varying

models for aircraft. J Guidance Control Dyn 2004;27(2):21828.

23.

Huang YG, McColl

WF. An improved simplex
method for
function
minimization. Comput-Aided
Des Arch 2006;4(2):718.

Shin JY, Balas GJ, Kaya MA. Blending methodology of linear Yue Ting is a postdoctor at
School of Aeronautic Science
parameter varying control synthesis of F-16 aircraft system. J
and
Engi-neering,
Beihang
Guidance Control Dyn 2002;25(6):10408.
University. He received the B.S.,
21. Stevens BL, Lewis FL. Aircraft control and simulation. 2nd ed.M.S. and Ph.D. degrees in
New York: John Wiley & Sons, Inc.; 2003.
20.

aircraft design from Beihang

University in 2004, 2006 and
2010 respectively. His main
research interest is aircraft flight
dynamics and control.
Wang Lixin is a professor in
Beihang University. His main
research interest lies in aircraft
design, flight dynamics and flight
control.
Ai Junqiang is a researcher in
The First Aircraft Design
Institute of AVIC. His main
research interest lies in aircraft
design.