Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

2019 Sixth Indian Control Conference (ICC)

December 18-20, 2019. IIT Hyderabad, India

Application of Generic Flight Controller Design Approach for A


Delta Canard Fighter Aircraft -ADMIRE
P Lathasree1, Abhay A Pashilkar2
and
N Sundararajan3
the fully movable nose wings. The flaps and elevons are

Abstract— This paper presents the application of the recently


developed Generic Flight Controller design approach by the located on the wing. The layout of ADMIRE control surfaces
authors for a delta canard fighter aircraft referred to Aero Data is shown in Fig. 2. The ability to generate rolling, pitching and
Model in Research Environment (ADMIRE). The generic flight yawing moments will be determined by the aerodynamic
controller developed for high performance fixed wing aircrafts characteristics of these control surfaces.
uses the good features of nonlinear dynamic inversion with time
scale separation, control allocation and integrator backstepping ADMIRE has a simple flight control system catering for
with the major advantage that the whole design cycle can be longitudinal and lateral-directional channels with basic
carried out quickly. The results obtained for ADMIRE using stability and handling quality characteristics [2, 4]. The
generic flight controller results have been compared with those longitudinal controller of ADMIRE is with pitch rate
obtained from the ADMIRE controller taken from literature. command following at low speeds and load factor command
The quick adoption of generic flight controller to ADMIRE has following at higher speeds. The lateral controller controls the
been demonstrated along with encouraging results. wind vector roll rate while directional control is used to
control the angle of sideslip.
I. INTRODUCTION
Any unstable aircraft augmented with quickly designed
viable full envelope flight controller can be used for real time
flying during the early design phase itself for sub-system
analysis, design and evaluation of avionics displays and to
understand aircraft behaviour. Development of a generic flight
controller for high performance aircraft was presented by the
authors in [1]. In the present paper, the authors demonstrated
the quick adaptability of generic flight controller to a delta
canard fighter aircraft referred to as the Aero Data Model in
Research Environment (ADMIRE). Performance of the
generic flight controller is demonstrated by comparing the
ADMIRE closed loop simulation responses obtained using
generic flight controller to those available in open literature. Figure 1. Approximate ADMIRE flight envelope for controller design
ADMIRE is a complete simulation model based on Generic
Aero data Model (GAM) and implemented in
Elevon
MATLAB/SIMULINK environment [2]. GAM is an open
source aerodynamic model corresponding to a single engine Rudder
fighter aircraft with delta canard configuration and valid for Canard
Mach number up to 2.5 and altitude 20000m [3].
In ADMIRE, GAM is augmented with the models of
engine, actuator and sensors. The flight envelope of ADMIRE
is limited to Mach number less than 1.2 and altitude below
6000 m. The flight envelope of ADMIRE is smaller than Leading edge flap
GAM since it is constrained due to the engine model. Fig. 1
shows the approximate flight envelope of ADMIRE. The Figure 2. Layout of ADMIRE control configuration
aircraft model can be trimmed and linearized within this
envelope. ADMIRE has a pair of canards (DRC and DLC i.e., The aircraft configuration data (geometry, mass, cg and
right and left canards) and four elevons (right outboard inertia) details are presented in Table 1.
(ROE), right inboard (RIE), left inboard (LIE) and left
outboard (LOE)) and rudder (DRUD) for control. Canards are

1 P Lathasree, Senior Principal Scientist, FMC Division, CSIR-National 3 N Sundararajan, School of Electrical & Electronics Engg., Nanyang
Aerospace Laboratories, Bangalore, 560017 e-mail: platha@nal.res.in Technical University, Singapore, e-mail: ensundara@ntu.edu.sg
2 Abhay A Pashilkar, Head, Systems Engineering Division, CSIR-
National Aerospace Laboratories, Bangalore-560017. e-mail:
apash@nal.res.in.

978-1-7281-3860-2/19/$31.00 ©2019 IEEE 479

Authorized licensed use limited to: Politecnico di Milano. Downloaded on February 18,2023 at 06:45:18 UTC from IEEE Xplore. Restrictions apply.
Typically, high performance fighter aircrafts are designed
TABLE 1. ADMIRE NOMINAL CONFIGURATION DATA to be unstable in the longitudinal axis to achieve the desired
Description Value agility. Designer must verify the stabilizability of the aircraft
Mass (Kg) 9100 within the operational AoA range, at each point within the
flight envelope shown in Fig. 1, for the chosen hardware
XCG (m) 0.0 standard [7]. The worst case pitch instability (TTD or Time
To Double) for the aircraft is shown in Fig. 4. It may be seen
YCG (m) 0.0
that the worst time to double occurs near the Mach number of
ZCG (m) -0.15 1.0. The operational constraints shown in Figs. 1, 3 and 4 are
typical for high performance fixed wing aircraft.
IX (kg-m2) 21000.0
2
After establishing the operational constraints, the fighter
IY (kg-m ) 81000.0 aircraft performance within the flight envelope will be
IZ (kg-m2) 101000.0 determined. Fig. 5 shows the achievable turn rate (𝑑𝜓 𝑑𝑡
) as a
function of Mach number. The maximum acheivable turn rate
IXZ (kg-m2) 2100.0 of 33.8 deg/sec is marked in Fig. 5. The operational
Wing Area (m ) 2
45.0 constraints and the performance achievable within these
constraints in the context of flight controller were discussed
Wing Span (b) (m) 5.2 till now. The operational constraints and performance have
been decided by the aircraft open loop characteristics.
Mean Aerodynamic 10.0
Chord (cbar) (m)

As seen from Fig. 1, operation of a fighter aircraft at the


low speed is limited by the maximum angle of attack. The open
loop stability and control within this maximum angle of attack
is found to be adequate. At the low speed limit, the aircraft
control surfaces are required to generate a minimum pitch
down acceleration of -0.3rad/sec2 at 1g [5]. The modified
Weissmann criteria [6] is used to arrive at the low speed limit.
The Angle of Attack (AoA, Alpha, α) – Mach number
envelope is shown in Fig. 3. It is seen that the maximum AoA
Figure 5. Turn rate capability.
is limited to 28 degrees. The second plot in Fig. 3 shows the
normal acceleration (Nz) as a function of Mach number. At Rest of the paper is organized as follows. Section II
higher speeds, the maximum Nz limits the operational AoA. presents the existing ADMIRE controller used for comparison.
The Nz is restricted between -3g to 9g over the flight envelope. Section III discusses the application of generic flight controller
The final plot in Fig. 3 shows the angle of attack as a function design approach to ADMIRE. Section IV presents the
of the equivalent airspeed (VEAS). The unit of VEAS is comparison of closed loop simulation results obtained with
Kilometer per hour (Kmph). The use of VEAS makes the AoA generic flight controller and existing ADMIRE flight
nearly independent of altitude. controller. Section V summarizes the conclusions from the
study.

II. EXISTING ADMIRE FLIGHT CONTROLLER


As mentioned in the previous section, the longitudinal
flight controller provides pitch rate and Nz command
following. The lateral controller provides roll control about
aircraft velocity vector and angle of sideslip control. The time
delays arise due to the physical implementation of the control
laws in an onboard computer and are modeled as transport
Figure 3. Operational Constraints. delays of 20 milli seconds and have been added to the
actuators. The time delay is modeled as a second order Pade
approximation. The actuator model used for four elevons, two
canards and rudder is a first order transfer function i.e.,
1
.
0.05𝑠+1

Sensor models used by the Flight Control System (FCS) are


incorporated in the model. Different models are considered
for air data sensors (VT (true air speed), α, , h (altitude)),
inertial sensors (pb , qb , rb , Nz) and attitude sensors (, ).
Figure 4. Pitch Instability Level (time to double)

480

Authorized licensed use limited to: Politecnico di Milano. Downloaded on February 18,2023 at 06:45:18 UTC from IEEE Xplore. Restrictions apply.
1 The SNDI controller structure has two outer loops with AoA
Air data sensors:
1+0.02s
and AoSS commanding the inner loop pitch rate and stability
1+0.005346𝑠+0.0001903𝑠 2
axis yaw rate respectively. The inner loop of this cascaded
Inertial Sensors: controller consists of separate feedbacks of pitch rate, stability
1+0.03082s+0.0004942s2
axis roll and yaw rates.
1
Attitude Sensors: Left & Right Canards
Left , Right Inboard and
1+0.0323s+0.00104s2 qcmd Outboard elevons
pscmd Rudder
acmd rscmd Aircraft
Outer Inner Control
+
ADMIRE model implementation assumed International cmd
Loop Loop Allocation
Actuator

Standard Atmosphere for trimming and simulation. A control


q
selector is used to distribute the longitudinal, lateral and ps
directional outputs to the seven control actuators (2 canards, Fast Variables rs
a
4 elevons and rudder) used by the FCS. All FCS gains are 
Slow Variables
scheduled with the altitude and Mach number. Fig. 6a shows
the top level schematic of ADMIRE controller implemented
Figure 7. Schematic of SNDI flight controller.
in Matlab/Simulink [2, 4]. Fig. 6b shows the expanded
Control System block of Fig. 6a depicting longitudinal and
A. Development of Generic Flight Controller with
lateral controllers. Simplified Nonlinear Dynamic Inversion (SNDI)
Approach
Coupled linear model considering longitudinal and
lateral-directional interaction is used for the inner loop
controller design. The inner loop with significant inertia
coupling terms is expressed in state-space form as:

((Iz − Ix )/Iy ) pr
ẋ = ((Iy − Iz )/Ix ) qr + Ay + Bu (1)

[((Ix − Iy )/Iz ) pq]


Figure 6a. Schematic of ADMIRE flight controller [2, 4] where,
x = [q p r]T (2)
y = [α q β p r]T (3)
u = [DRC DLC ROE RIE LIE LOE DRUD] (4)
It may be noted that we have deviated from standard state
space representation in (1). The primary aerodynamic
dependence on AoA and AoSS has been introduced into the
equation, in addition to the dependence on angular rates. The
matrix A is termed as augmented plant matrix since it is
different from standard plant matrix.
For the decoupled control of each of the three rotational
axes, three pseudo-controls are defined:
Figure 6b. Schematic of ADMIRE flight controller [2, 4]
ū = [δpitch δroll δyaw ]T (5)
III. ADMIRE WITH GENERIC FLIGHT CONTROLLER
Further, the body axis roll and yaw rates are transformed
The development of Generic Flight Controller with into the stability axis rates. The transformed 𝑥 and 𝑦 vectors
Simplified Nonlinear Dynamic Inversion (SNDI) approach are given by:
has been discussed in detail in [1, 8, 9]. However, for
completeness sake it is briefly presented here. The SNDI x̄ = [q ps rs ]T (6)
approach uses the good features of nonlinear dynamic ȳ = [α q β ps rs ]T (7)
inversion with time scale separation, control allocation and
integrator backstepping. The generic flight controller is Equations (8-10) show the transformation of original
designed to be angle of attack, stability axis roll rate (ps ) and variables 𝑥, 𝑦 and 𝑢 to 𝑥̄ , 𝑦̄ and 𝑢̄ respectively:
angle of sideslip (AoSS) command following for the pitch, 1 01×2
roll and yaw channels. The pitch stick, roll stick and rudder x=[ ] x̄ = T1 x̄ (8)
02×1 Ts−1
pedal commands are shaped appropriately and scaled to result
I3×3 03×2
in AoA, stability axis roll rate and AoSS () commands. y=[ ] ȳ = T2 ȳ (9)
Block schematic of the SNDI controller is shown in Fig. 7. 02×3 Ts−1

481

Authorized licensed use limited to: Politecnico di Milano. Downloaded on February 18,2023 at 06:45:18 UTC from IEEE Xplore. Restrictions apply.
u = Sū (10)  If the optimization finds a solution greater in
Here, S is the control allocation matrix. The size of T1 magnitude than 1.2, it is reset to 1.2. The maximum
matrix is 3 by 3 and size of T2 matrix is 5 by 5. The variable differential deflection of elevons is 25 degrees and
Ts in the above equations represents the rotational rudder is 30 degrees. Limiting KR5 gain to 1.2
transformation matrix which transforms the body axis rates prevents rudder surface saturation due to roll
(𝑝, 𝑟) to the stability axis (𝑝𝑠 , 𝑟𝑠 ) as shown in (11): command at the expense of some sideslip buildup
𝑝𝑠 during roll about velocity vector.
𝑐𝑜𝑠 𝛼 𝑠𝑖𝑛 𝛼 𝑝 𝑝
[𝑟 ] = [ ] [ ] = 𝑇𝑠 [ ] (11)  The maximum roll rate in the stability axis (velocity
𝑠 − 𝑠𝑖𝑛 𝛼 𝑐𝑜𝑠 𝛼 𝑟 𝑟
vector roll rate) can be computed using the control
allocation matrix (S) and used for the forward path
With the above transformations, the simplified equations
command scaling throughout the airspeed range.
for the rotational dynamics can be written as:
C. Handling Qualities (HQ) Criteria
((Iz − Ix )/Iy ) pr During the flight controller design, to meet HQ criteria,
following aspects are considered. The stick path design has
T1 x̄̇ = ((Iy − Iz )/Ix ) qr + AT2 ȳ + BSū (12)
been carried out accordingly.
[((Ix − Iy )/Iz ) pq]  The time derivative of AoA (α̇ ) should not exceed
25deg/sec through out the envelope.
Equation (12) can be rearranged as:  At each flight condition, the roll rate to be demanded is
determined as the maximum achievable full stick roll rate
with a maximum of 200 deg/sec.
((Iz − Ix )/Iy ) pr
 Maximum roll acceleration for full stick should be less
(BS)−1 T1 x̄̇ = (BS)−1 ((Iy − Iz )/Ix ) qr (13) than 600 deg/sec/sec.
 AoSS build-up during roll maneuvers should be
[((Ix − Iy )/Iz ) pq] minimum possible.
+(BS)−1 AT2 ȳ + ū
D. Inner Loop Design
where (BS)−1 is the inverse of matrix product (BS). The feedback gains (Kq, Kps and Krs) of inner loop
B. Control Allocation corresponding to pitch, roll and yaw channels are chosen such
that the time scale separation between actuator bandwidth and
ADMIRE has seven control surfaces. The canards and inner loop bandwidth ensures no rate limiting during
elevons could be used together or in differential mode. The maneuvers. The inner loop feedback gains are scheduled with
pitch, roll and yaw channels are to be controlled using these dynamic pressure.
seven control surfaces. Hence, the transformation matrix has
dimension of 7 by 3. The control matrix B of dimension E. Outer Loop Design
3 by 7 is studied to determine the structure of the control For the outer loop controller design, dynamics of angle of
allocation matrix, S. The following approach is followed to attack (α) and sideslip angle (β) are expressed as functions of
arrive at the control allocation matrix: accelerations and flight path angles (μ, γ).
 For pitch axis control, the together canards (KP1) and g
α̇ ≅ q − (Nz -CosμCosγ) − ps Tanβ (15)
together elevons (KP3) are used. For roll axis control, V
g
the differential canards (KR2), and differential elevons β̇ ≅ −rs + (SinμCosγ − Ny ) (16)
V
(KR4) are used. Finally for yaw axis control,
differential canards (KY2), differential elevons (KY4) The outer loop control law based on above
and rudder (KY5) are used. approximations is given by:
The control allocation matrix is given by: g
qcmd = K α (αcmd − α) + (Nz − CosμCosγ) (17)
V
KP1 −KR 2 −KY2
+ps Tanβ
KP1 KR 2 KY2
g
KP3 −KR 4 −KY4 rcmd = −K β (βref − β) + (SinμCosγ − Ny ) (18)
V
S = KP3 −KR 4 −KY4 (14)
KP3 KR 4 KY4 Here g is acceleration due to gravity, V is the true air speed
KP3 KR 4 KY4 and Ny is the lateral acceleration. The gains are chosen to be:
K α = 2.5rad/s/rad and K β = 2.5rad/s/rad and are constant.
[KP5 KR 5 KY5 ]
 The gain KR5 (aileron to rudder interconnect gain) and IV. COMPARSION OF SIMULATION RESULTS
maximum of KY2, KY4 or KY5 is chosen for
The approaches used by existing ADMIRE controller and
optimization, such that (BS)−1 T1 is a diagonal matrix
the proposed generic flight controller are given in Table 2.
resulting in the control decoupling of pitch, roll and
yaw axes.

482

Authorized licensed use limited to: Politecnico di Milano. Downloaded on February 18,2023 at 06:45:18 UTC from IEEE Xplore. Restrictions apply.
comparison of aircraft responses is shown in Fig. 8 for full
TABLE 2. COMPARISON OF CONTROLLERS pitch stick step input (pstick). The gravity correction terms
Sl. ADMIRE Controller Generic Flight corresponding to pitch and yaw channels are present in
No Controller (17, 18) with respect to Generic Flight Controller.
ADMIRE controller Generic Flight Comparison of closed loop aircraft responses for pitch stick
contains longitudinal Controller has AoA, step input at Mach number of 0.5 and altitude of 15000ft are
and lateral part. Stability axis roll rate shown in Fig. 9. The AoA from Generic Flight Controller is
Longitudinal controller and beta command found to be more ( 3deg) due to the addition of gravity
1 provides pitch rate/ Nz following for pitch, correction terms for pitch channel. Corresponding control
command following roll and yaw surface deflections are shown in Fig. 10. It is observed that
and lateral controller channels the flight controller primarily uses the elevons with smaller
provides roll control respectively. deflections of canards for the pull up maneuver as elevons are
the most effective control surfaces for this maneuver.
around the velocity
vector of the aircraft
and angle of sideslip
control
Control selection Control mixing is
scheme is used to used for control
distribute the output of t power utilization
2 control channels, to the among the seven
seven control actuators control surfaces.
used by the FCS. Control allocation
Control selector is matrix is scheduled
scheduled with Mach with dynamic
number and altitude. pressure Figure 8. Responses of aircraft for maximum pitch stick step input at
For ADMIRE flight The inner loop Mach number 0.5 and altitude of 15000ft (gravity correction and inertia
controller, design is coupling terms are not included in Generic Flight Controller).
feedback gains are
3 carried out at many computed as inverse
points in the envelope. function of their
respective
dimensional control
derivatives. Hence,
there will be minimal
design effort.
Kinematic coupling, Stability axis roll and
gravity correction and yaw rate feedback is
4 inertia coupling are not used to address
addressed. kinematic coupling.
Gravity Correction
terms and Inertia Figure 9. Responses of aircraft for pitch stick step input at Mach number 0.5
and altitude of 15000ft.
coupling terms are
added.
Longitudinal control is This approach allows
split into pitch rate us to quickly design a
control till Mach viable baseline
5 number 0.58 and Load nonlinear controller
factor control above for the entire flight
Mach number 0.62. For envelope
the region in between a
blending function is
used

Comparison of responses obtained using ADMIRE flight


controller and the generic flight controller has been made. To Figure 10. Responses of control surfaces for pitch stick step input at Mach
demonstrate better match, gravity correction and inertia Number 0.5 and altitude of 15000ft `
coupling terms are not included in generic flight controller as
they are not considered in ADMIRE controller. Corresponding

483

Authorized licensed use limited to: Politecnico di Milano. Downloaded on February 18,2023 at 06:45:18 UTC from IEEE Xplore. Restrictions apply.
The closed loop aircraft response and the control surface
deflections are shown in Figs. 11 and 12 respectively for a
doublet roll stick input (rstick) at Mach Number of 0.5 and
altitude of 15000ft. It is seen that the aircraft is able to achieve
a roll rate of about 200deg/s. The Ny, AoA and AoSS
excursions are kept small during this roll rate indicating good
turn coordination. It is also observed that the canards and
elevons are used in the differential mode while the rudder is
used to suppress the sideslip. Response to coupled full pitch
and roll stick inputs at Mach number 1.0 and altitude 15000ft.
is shown in Fig. 13. It may be noted that AoSS, Ny responses Figure 14. Responses of aircraft for pitch stick input and discrete gust
from generic flight controller are very small. Aircraft injected at 5sec at Mach number 1.0 and altitude 15000 ft.
responses to discrete gust input inject at 5sec along with pitch
stick step input at Mach number 1.0 are shown in Fig. 14. V. CONCLUSION
The design process for the ADMIRE aircraft followed a
typical approach. First, the open loop performance, stability
and control parameters for the aircraft were determined.
Secondly, based on this, the region where it was possible to
satisfy the low speed and high speed performance limits while
maintaining acceptable stability and control parameters as per
MIL specifications were established. The open loop study
also indicated the optimal mix of the flight controls to be used
to achieve the performance. Finally, the closed loop gains
were selected by SNDI approach keeping in view the handling
quality requirements, actuator bandwidth, actuator rate and
control surface position limits for large amplitude maneuvers.
Figure 11. Responses of aircraft for roll stick doublet input at Mach
Number 0.5 and altitude of 15000ft.
The entire process could be completed in about three weeks
in a more or less sequential manner. The comparison of the
responses of the generic flight controller against the flight
controller provided with the ADMIRE model demonstrated
similar performance and in some cases better decoupling. The
authors believe that such an orderly process prevents iterative
back and forth and is hence less time consuming compared to
other methods of design.

REFERENCES
[1] P. Lathasree, A. A. Pashilkar and N. Sundararajan, “Generic Flight
Controller design for a high performance fixed wing aircraft,” Indian
Control Conference,9 – 11 January, 2019, IIT Delhi, New Delhi, India
[2] Lars Forssell and Ulrik Nilsson, “ADMIRE The Aero Data Model in
Research Environment Version 4.0, Model Description”, Report No.
Figure 12. Responses of control surfaces for roll doublet stick input at FOI-R--1624--SE, FOI Swedish Defence Research Agency, Systems
Mach Number 0.5 and altitude of 15000ft. Technology, SE -164 90, Stockholm, 2005
[3] Hans Backstrom and Erik Kullberg, “Report on the usage of Generic
Aerodata Model”, 1996
[4] Declan Bates and Martin Hagstrom (Eds.), Nonlinear Analysis and
Synthesis Techniques for Aircraft Control. Lecture Notes in Control
and Information Sciences (365). Springer - Verlag Publications, 2007,
pp. 35-68
[5] Anonymous, Flying Qualities of Piloted vehicles, MIL-HDBK-1797,
1997.
[6] W. H. Mason, High Angle of Attack Aerodynamics downloaded 2018,
see http://www.aoe.vt.edu/~mason/Mason_f/
ConfigAeroHiAlphaNotes.pdf.
[7] P. S. Rao, G. K. Singh, G. S. Deodhare, and S. Chetty, “Pitch
stabilizability of an unstable aircraft”, AIAA-2003-5411, AIAA
Atmospheric Flight Mechanics Conf., 11-14 August, 2003, Austin,
Texas.
[8] P. Lathasree, A. A. Pashilkar and N. Sundararajan, “Fast nonlinear
Figure 13. Responses of aircraft for maximum pitch and roll stick inputs at flight controller design for high performance fighter aircraft,” Indian
Mach Number 1.0 and altitude of 15000ft. Control Conference,5 – 7 January, 2015, IIT Madras, Chennai, India
[9] P. Lathasree, S. Ismail, and A. A. Pashilkar, “Design of nonlinear flight
controller for fighter aircraft,” Advances in Control and Optimization
of Dynamical Systems Conference, 13-15 March, 2014, IIT Kanpur,
Kanpur, India.

484

Authorized licensed use limited to: Politecnico di Milano. Downloaded on February 18,2023 at 06:45:18 UTC from IEEE Xplore. Restrictions apply.

You might also like