You are on page 1of 6

2020 International Conference on Power, Instrumentation, Control and Computing (PICC)

High Alpha Maneuvering with a Laterally


Asymmetric Fighter Aircraft: A Backstepping Control
Approach
2020 International Conference on Power, Instrumentation, Control and Computing (PICC) | 978-1-7281-7590-4/20/$31.00 ©2020 IEEE | DOI: 10.1109/PICC51425.2020.9362423

Anukaran Khanna Bijoy K. Mukherjee


Dept. of Electrical and Electronics Engg. Dept. of Electrical and Electronics Engg.
Birla Institute of Technology and Science Birla Institute of Technology and Science
Pilani, India Pilani, India
anukaran.bits@gmail.com bijoy.mukherjee@pilani.bits-pilani.ac.in

Abstract— Asymmetric release of payload or partial damage of


consumption or partial structural damage. The asymmetry,
wing shifts the center-of-gravity (c.g.) of a fighter aircraft to new predominantly in the lateral dynamics leads to coupling of
positions resulting in a highly coupled and nonlinear asymmetric longitudinal and lateral-directional dynamics and therefore, to
dynamics. Since controlling the aircraft becomes much more increased nonlinearity and complexity. The situation further
challenging when it tries to perform some high alpha maneuver aggravates when the aircraft has to execute some high alpha or
with such lateral asymmetry, implementation of nonlinear control
high angle-of-attack maneuver under such asymmetry posing
becomes mandatory for the safety of the aircraft. In the present
paper, the highly coupled asymmetric dynamics is first converted significant challenges for the control engineers. Though some
to the strict feedback form and thereafter a backstepping control works for asymmetric dynamics are reported in the literature,
is designed to autonomously execute the high alpha cobra they mostly consider steady level flight conditions for civil
maneuver under significant lateral c.g. shift. The simulation aircraft [2-5]; works on asymmetric fighter aircraft performing
results show the maneuver performance under lateral asymmetry some demanding maneuver is scarce [6].
to be almost the same as that without any asymmetry.
The present work proposes a backstepping based nonlinear
Index Terms— asymmetric aircraft, backstepping control, cobra control to execute a high alpha maneuver under lateral
maneuver, F-18 HARV, high-alpha.
asymmetry arising from unbalanced ejection of stores [7-9] and
Nomenclature— uncertainty in the aerodynamic coefficients. Since the dynamics
u,v,w velocity component in body axes is extremely nonlinear, a backstepping controller is considered.
p,q,r roll, pitch, and yaw body rates To our best knowledge, backstepping control design for
φ,θ,ψ roll, pitch, and yaw Euler angles asymmetric aircraft performing some demanding maneuver is a
α angle of attack novel attempt not reported in the literature. Although, sliding
β sideslip angle
mode control has been attempted in the recent literature, SMC
CX,CY,CZ forward, side, and normal aerodynamic force
coefficient is known to suffer from the chattering problem. In the present
Cl,Cm,Cm rolling, pitching, and yawing aerodynamic work first the nominal/symmetric dynamics is considered and
moment coefficient then the design method is extended to the asymmetric case, for
TX,TY,TZ thrust component along body axes which the coupled 6-DOF aircraft dynamics is cast into the strict
MTX,MTY,MTZ moment due to thrust along three body axes feedback form as required for backstepping control design.
xcm,ycm,zcm position of c.g. of asymmetric aircraft from Cobra maneuver is considered for validation of the proposed
nominal c.g. location
control scheme. In Cobra maneuver, the aircraft first pitches up
I. I NTRODUCTION to 900or more and then quickly pitches down to initial value
within few seconds [10]. The ability to alter the pitching angle
Modern day fighter aircraft are designed in such a manner within few seconds categorizes this maneuver as one of the most
that, they can carry heavy payloads with them and able to valuable maneuver in dog fight situations. It is deemed as
execute various maneuver under critical combat conditions. To standard maneuver to assess the maneuverability and agility of
keep symmetry of the aircraft intact it is a common practice to any modern-day fighter aircraft. The F-18 High Alpha Research
release payloads/dummies in pairs. However, sometimes stores Vehicle (HARV) is considered for simulation and validation
may not be available in identical pairs or there may be non- purposes as its high alpha aerodynamic database is available in
paired release due to malfunction in the ejection mechanism. the open domain [11]. The aircraft is assumed to have elevator,
Under such conditions aircraft becomes asymmetric and its aileron, rudder, pitch thrust vector and yaw thrust vector as the
center-of-gravity (c.g.) shifts to a new lateral position control surfaces and all of them are position and rate constrained
depending upon the mass ejected and its distance from the body with a first order stable dynamics.
center line [1]. Therefore, it is expected from modern day fighter
aircraft to successfully execute various offensive or defensive The ensuing sections of the paper are organized as follows.
maneuvers even under considerable lateral mass asymmetry. It Section II describes the control formulation for symmetric
may be noted that apart from asymmetric release of stores, aircraft in detail followed by simulation results. Section III
lateral asymmetry may also result from unbalanced fuel presents the control formulation for the asymmetric aircraft and
compares the simulation results of the symmetric case.
978-1-7281-7590-4/20/$31.00
Authorized licensed use limited to: BIRLA INSTITUTE OF TECHNOLOGY ©2020
AND SCIENCE. Downloaded IEEE
on August 10,2021 at 07:00:52 UTC from IEEE Xplore. Restrictions apply.
2020 International Conference on Power, Instrumentation, Control and Computing (PICC)

Fig.1. Block Diagram of the Closed Loop Control Scheme

Finally, conclusions are presented in section IV.  a 


 
M x   qSbCl a qSbCl e qSbCl r 0 l z  T T    e 
   
II.B ACKSTEPPING CONTROL FOR S YMMETRIC D YNAMICS M y    0 qScCm e 0 l x  T T 0   r 
 
M   qSbCn a qSbCn e qSbCn r 0 l xT T   ptv 
First we develop a controller for the symmetric aircraft. In order  z c 
 
to execute the cobra maneuver, Euler angles roll and yaw are  ytv 
directed to zero for the entire time span of the maneuver, whereas [Mx;My;Mz]s and [Mx;My;Mz]c are part of moment vector
a bell shaped curve with peak value of 900 is considered for the dependent upon state and control respectively.
desired pitch angle profile. The objective of the controller is to
make the state variables [ ,  , ] track the given desired Let x1  [ , , ] , x2  [ p, q, r ] and u  I 1[ M x M y M z ]Tc . So
trajectories [ d ,  d , d ] . To successfully execute the maneuver, that [ M x M y M z ]Tc  I .u .With the choices of x1 , x2 and u , (1)
controller first calculates the moments about the three body axes.
and (2) reduces to the strict feedback form as follows:
The computed moments are then distributed among the available
control surfaces via pseudoinverse control allocation algorithm. x1  f1 ( x ) x2 (3)
Throttle is commanded in an open loop mode. It is assumed that
all the actuators possess first order dynamics and position and x2  f 2 ( x )  u (4)
rate limits as given in Table I. The block diagram of the control
scheme is shown in Fig. 1. Let us introduce the error state variables z1 and z2 as
The rotational kinematics and dynamics in body fixed frame follows:
[12] for the aircraft can be rearranged as, z1  x1  x1d (5)
   1 sin  tan  cos  tan    p 
  
    0 cos   sin    q  (1) z 2  x2  x2d (6)
 
   0 sin  sec  cos  sec    r 
where x1d and x2d are the desired trajectories of x1 and x2 .
 p   ( I yy  I zz )qr  I xz pq 
 q   I  ( I  I ) pr  I ( r 2  p 2 )   The dynamic equations of the error states are given as follows:
     zz
 1
xx xz 
 r   ( I  I ) pq  I qr  z1  x1  x1d  f1 ( x ) x2  x1d (7)
 xx yy xz

M x  M x  To improve steady state performance, an integral term is
1   1   added to the standard Lyapunov function for the x1 dynamics
 I  M y   I  M y  (2) assuming x2 to be the virtual control as
M  M 
 z s  z c 1 T 1
V1  z1 z1  p1T p1 (8)
2 2
where, [I] is the inertia matrix,

 pb rb 

where p1  K12 z1dt with K12 being a constant symmetric

 qSb(Cl    Clp 2V  Clr 2V )  positive definite matrix.


M x   
  qc V1  z1T z1  p1T p1
M y   qSc (Cm  Cmq ) 
 2V 
M    = z1T [ f1 x2  x1d ]  p1T K12 z1
 z s  qSb(C   C pb rb 
n np  C nr )
 2V 2V  Let us choose f1 x2  x1d   K11 z1  K12 p1 (9)

where K11 is a constant positive definite matrix.


Authorized licensed use limited to: BIRLA INSTITUTE OF TECHNOLOGY AND SCIENCE. Downloaded on August 10,2021 at 07:00:52 UTC from IEEE Xplore. Restrictions apply.
2020 International Conference on Power, Instrumentation, Control and Computing (PICC)

 V1  z1T (  K11 z1  K12 p1 )  p1T p1 Position Rate limit Bandwidth
limit
  z1T K11 z1  z1T K12 p1  p1T K12 z1
Aileron (δa) ±35 deg ±100 deg/s 10.0 Hz
  z1T K11 z1 0
Elevator (δe) + 10 deg, ± 40 deg/s 10.0 Hz
Now, from (9), f1 x2   K11 z1  K12 p1  x1d -25deg
From here, the desired value of x2 is obtained as Rudder (δr) ±30 deg ±82 deg/s 10.0 Hz

 x2 d  ( f1 ) 1 (  K11 z1  K12 
2
z1dt  x1d ) (10) Pitch Nozzle ±20 deg ±80 deg/s 10.0 Hz
(δptv)
The Lyapunov function for the complete dynamics is chosen Yaw Nozzle ±20 deg ±80 deg/s 10.0 Hz
as (δytv)
1 T 1 1 1 TABLE II: Controller Parameters
V2  z1 z1  p1T p1  z2T z2  p2T p2 (11)
2 2 2 2 0.1 0 0
K11
0 6.75 0

where p2  K 22 z2 dt , K22 being a constant symmetric
0 0 0.5
positive definite matrix. K12 0.1 0 0
0 0.1 0
V2  z1T z1  p1T p1  z T2 z2  p2T p 2
0 0 0.1
= z1T [ f1 x2  x1d ]  p1T K12 z1  z2T z2  p2T K 22 z2 K21 0.3 0 0
0 0.7 0
= z1T [ f1 z 2  f1 x2 d  x1d ]  p1T K12 z1  z2T z2  p2T K 22 z 2 0 0 0.1
K22 0.1 0 0
Substituting (10) in the above, 0 0.1 0
V2  z1T [ f1 z2  K11 z1  K12 p1  x1d  x1d ]  0 0 0.1
150m/s. Figure 2 depicts the state and control time profiles for
p1T K12 z1  z2T z2  p2T K 22 z2 the maneuver. From Fig.2. it can be seen that good tracking
result is obtained as the actual and commanded pitch angle
  z1T K11 z1  z1T f1 z2  z1T K12 p1  p1T K12 z1  profiles are nearly overlapping; the roll angle, yaw angle, and
sideslip angle are almost zero throughout. Over 50% drop in
z2T z2  p2T K 22 z2
velocity is achieved while about 200m altitude is gained after
the maneuver. The control surface deflections of aileron, rudder,
Let us choose z2   K 21 z 2  K 22 p2  f1T z1 (12)
and yaw nozzle did not show any considerable variations and
where K21 is a constant positive definite matrix therefore, are not included in Fig. 2.
Elevator and pitch nozzle deflections remain mostly within their
V2   z1T K11 z1  z1T f1 z2  z2T K 21 z2  z2T K 22 p2
saturation limits. Elevator is observed to hit its saturation limit
 z2T f1T z1  p2T K 22 z2 of +10 deg for a small duration when pitch angle is close to 900.
To enhance the power of thrust vector control, 50% throttle is
=  z1T K11 z1  z 2T K 21 z2  0 propelled in an open loop mode. Controller parameters as in (10)
and (13) are fixed after some trial and error and are shown in
Now from (12), Table II.


2
x2   K 21 z 2  K 22 z2 dt  f1T z1  x2 d


2
f 2  u   K 21 z 2  K 22 z2 dt  f1T z1  x2 d


2
u   f 2  K 21 z 2  K 22 z2 dt  f1T z1  x2 d (13)

Equation (13) gives the final control input where x2d is given
by (10).
For numerical simulation of cobra maneuver, a maneuver
duration of 6s is considered starting at t=5s. Before the
maneuver the aircraft is assumed to be trimmed at 2000m and
TABLE I: Saturation level and bandwidth of control effectors

Authorized licensed use limited to: BIRLA INSTITUTE OF TECHNOLOGY AND SCIENCE. Downloaded on August 10,2021 at 07:00:52 UTC from IEEE Xplore. Restrictions apply.
2020 International Conference on Power, Instrumentation, Control and Computing (PICC)

  qw  rv 
  ru  pw 
 
  pv  qu 
 
 I xz pq  ( I yy
  I zz
 ) qr  I xy  (q  r )  
 rp  I yz 2 2

 
 qr  ( I zz
 I xy   I xx
 )rp  I yz  (r 2  p 2 ) 
 pq  I xz
 
 rp  ( I xx
 I yz   I yy ) pq  I xz  ( p 2  q 2 ) 
 qr  I xy
 (q 2  r 2 ) xcm  pqycm  rpzcm 
 
 (r 2  p 2 ) ycm  pqxcm  qrzcm 
 2 2 
 ( p  q ) zcm  rpxcm  qrycm  
 m(qu  pv) y  m(ru  pw) z 
 cm cm 
 m( pv  qu ) xcm  m(rv  qw) zcm 
  
 m ( pw  ru ) xcm  m(qw  rv ) ycm 
 Fx 
 m 
  g sin    
   Fy 
 g cos  sin    m 
 g cos  cos    
    Fz  (14)
  zcm mg cos  sin   ycm mg cos  cos    m 
  z mg sin   x mg cos  cos    
 cm cm
 M x 
 ycm mg sin   xcm mg cos  sin    M y 
 
 M z 

where m' and [I'] are mass and inertia matrix of the aircraft
respectively after store ejection; and [Mx My Mz]T and [Fx Fy Fz]T
are external moment and force vectors acting on the aircraft
consisting of aerodynamic and propulsive components. It is
important to note here that like the gravitational force, there also
exist a moment due to gravity term as the c.g. of the aircraft is
now away from the origin of the body frame.
From the above nonlinear coupled equation, the rotational
dynamics part can be taken out as:
Fig.2. Cobra maneuver response in symmetric dynamics   qw  rv  (q 2  r 2 ) x  pqy  rpz 
 p  1  cm cm cm
    1 2 
 
q
    I 
m
[ A]    A    ru  pw  ( r 2
 p 2
) y cm  pqx cm  qrz cm  
 r     
III. B ACKSTEPPING CONTROL FOR A SYMMETRIC 2 2
   pv  qu  ( p  q ) zcm  rpxcm  qrycm 
D YNAMICS
 I xz
 pq  ( I yy  I zz  rp  I yz (q 2  r 2 )  m(qu  pv ) ycm  m(ru  pw) zcm 
 ) qr  I xy
 
The backstepping control formulation discussed in previous  I xy
 qr  ( I zz
  I xx  (r 2  p 2 )  m ( pv  qu ) xcm  m( rv  qw) zcm  
 )rp  I yz pq  I xz
 
section is extended to the c.g. shifted asymmetric aircraft in this  I  rp  ( I xx
  I yy
 ) pq  I xz  ( p 2  q 2 )  m( pw  ru ) xcm  m(qw  rv ) ycm 
 qr  I xy
 yz 
section. In the plane of symmetry, if the shifted c.g. position is
at [xcm ycm zcm]T from the nominal reference point, then the  Fx  M x   1
M x 
       (15)
 A   F y    M y     I   m  [ A ] 2 
1
force and moment balance equation in body reference frame is M y 
F       M 
given by [13-14].  z  s  M z  s   z c

 1 0 0 0 zcm  ycm   u  where, as considered in the previous section, [Mx My Mz]Tc and
 0 1 0  zcm 0 xcm   v  [Mx My Mz]Ts are part of external moment dependent upon control
  
 0 0 1 ycm  xcm 0   w  and state respectively; and
   I xx
  
  I xz
 0  mzcm mycm 
I xx 
 I xy    p  
 I xz 
 I xy

 0 mzcm mycm 
   
 mz I     I xy
   ; A   mzcm
I yy  I yz 0 mxcm 
0  mxcm 
 I xy I yy  I yz   q 
 cm      mycm mxcm 0 
  mycm   I yz I zz
  I xz  
 mxcm 0 
 I xz 
 I yz    r 
I zz
The rotational kinematic equation remains unchanged under c.g.
asymmetry as given by

Authorized licensed use limited to: BIRLA INSTITUTE OF TECHNOLOGY AND SCIENCE. Downloaded on August 10,2021 at 07:00:52 UTC from IEEE Xplore. Restrictions apply.
2020 International Conference on Power, Instrumentation, Control and Computing (PICC)

   1 sin  tan  cos  tan    p 


  
    0 cos   sin    q  (16)
 
   0 sin  sec  cos  sec    r 
It is easy to observe that (15) and (16) are in the same form
as (3) and (4) with the f2 term being more complex. Hence, the
control design method and stability proof remain similar.
To perform numerical simulation in case of asymmetric
condition, it is assumed that the aircraft is flying at an altitude
of 2000m with velocity of 150m/s as in the symmetric case.
Further, the aircraft is assumed to carry a single store of 800 Kg
housed underneath the starboard wing. The store is positioned
laterally at one-third of the semi-span which is about 1.9m and
at 0.45m vertically downward from the nominal reference point
in the plane of symmetry causing a lateral c.g. shift of 12.06
cm.

When the performance of the nominal backstepping


controller as designed for the symmetric aircraft is examined
under the aforementioned c.g. variation, significant
deterioration in the maneuver performance is noticed as shown
in dotted plots in Fig. 3. However, when the controller is
redesigned considering (14) and (16), the maneuver
performance improves significantly and is found to be almost
identical to the nominal one, as evident from the firm plots in
Fig. 3.

Fig.3. Comparison of cobra maneuver response under lateral c.g. shift

IV.C ONCLUSION
Backstepping control design for automatically executing the
high alpha cobra maneuver was considered for the F18 HARV
aircraft under balanced as well as unbalanced lateral c.g.
locations. It was shown that the dynamics in either case can be
cast in the strict feedback form. It was further shown that the
controller designed based on the dynamics for the symmetric c.g.
position suffers from significant performance degradation under
lateral c.g. variation. When the control is modified based on the
asymmetric dynamics then the nominal performance was almost
recovered back. In both the cases, maneuver performances were
validated against a modest ±10% uncertainty in the aerodynamic
coefficients of the aircraft. Some robustness enhancement
schemes may be combined with the baseline backstepping
control to equip it to handle higher levels of uncertainty in the
aerodynamic database.

APPENDIX
Aircraft Data:
Mass (m): 11840 kg
Length (l): 17.06 m
Wingspan (b): 11.40 m
Authorized licensed use limited to: BIRLA INSTITUTE OF TECHNOLOGY AND SCIENCE. Downloaded on August 10,2021 at 07:00:52 UTC from IEEE Xplore. Restrictions apply.
2020 International Conference on Power, Instrumentation, Control and Computing (PICC)

Wing area (S): 37.16 m2


Mean chord ( c ): 3.51 m
Roll-axis inertia (Ixx): 3.089 × 104 kg m2
Pitch-axis inertia (Iyy): 2.396 × 105 kg m2
Yaw-axis inertia (Izz): 2.599 × 105 kg m2
X-Z axis Product of inertia (Ixz): -3.1243×103 kg m2

REFERENCES
[1]. Mukherjee, B.K., and Sinha, M., “Large Angle Maneuvering with an
Asymmetric Aircraft: A Single Loop Control Formulation”, Proceedings of the
Guidance, Navigation, and Control Conference, Florida, USA, 08-12 January
2018.
[2]. Bacon, B.J.,and Gregory, I.M. “General Equations of Motion for a Damaged
Asymmetric Aircraft”, Proceedings of the AIAA Atmospheric Flight Mechanics
Conference and Exhibits, South California, USA, 20-23 August 2007.
[3]. Nguyen, N., Krishnakumar, K., Kaneshige, J., and Nespeca, P., “Flight
Dynamics and Hybrid Adaptive Control of Damaged Aircraft”, Journal of
Guidance, Control, and Dynamics, Vol. 31, No. 5, 2008.
[4]. Liu, Y., Tao, G., and Joshi, S.M., “Modeling and Model Reference Adaptive
Control of Aircraft with Asymmetric Damage”, Journal of Guidance, Control, and
Dynamics, Vol. 33, No. 5, 2010.
[5]. Guo, J., Tao, G., and Liu, Y., “Multivariable Adaptive Control of NASA
Generic Transport Aircraft Model with Damage”, Journal of Guidance, Control,
and Dynamics, Vol. 34, No. 5, 2011.
[6]. Mukherjee, B.K., and Sinha, M., “A Single Loop Dynamic Inversion Control
for a Fighter Aircraft Executing Rapid Large Amplitude Maneuvers”, 14th IEEE
India Council International Conference, India, 15 - 17 December 2017.
[7]. Krtic, M., Kanellakopoulos, I., and Kokotovic, P., “Nonlinear and Adaptive
Control Design”, Wiley, New York, 1995.
[8]. Lee, T., and Kim, Y., “Nonlinear Adaptive Flight Control Using
Backstepping and Neural Networks Controller”, Journal of Guidance, Control,
and Dynamics, Vol. 24, No. 4, 2011.
[9]. Sadati, S.H., Parvar, M.S., Menhaj, M.B., and Bahrami, M., “Backstepping
Controller Design Using Neural Networks for a Fighter Aircraft”, European
Journal of Control, Vol.13, No.5, 2007.
[10]. Wang, N., Ma, R., Chang, X., and Zhang, L., “Numerical Virtual Flight
Simulation of Quasi-Cobra Maneuver of a Fighter Aircraft”, Journal of Aircraft,
Vol. 57, No. 4, 2020.
[11]. Khatri, A.K., and Sinha, N., “Aircraft Maneuver Design using Bifurcation
Analysis and Nonlinear Control Techniques”, 49th AIAA Aerospace Sciences
Meeting including the New Horizons Forum and Aerospace Exposition, Florida,
USA, 4 - 7 January 2011.
[12]. Stevens, B.L., and Lewis, F.L., “Aircraft Control and Simulation”, Wiley,
New York, 1992.
[13]. Mukherjee, B.K., and Sinha, M., “Extreme aircraft maneuver under sudden
lateral CG movement: Modeling and control”, Aerospace Science and
Technology, Vol.68, 2017.
[14]. Mukherjee, B.K., and Sinha, M., “Nonlinear dynamics and control of a
laterally mass varying fighter aircraft”, Journal of Aerospace Engineering, Vol.
232, No.16, 2018.

Authorized licensed use limited to: BIRLA INSTITUTE OF TECHNOLOGY AND SCIENCE. Downloaded on August 10,2021 at 07:00:52 UTC from IEEE Xplore. Restrictions apply.

You might also like