See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/312323342

Modeling and Simulation of Longitudinal Autopilot for General Aviation

Aircraft

Conference Paper · May 2016

DOI: 10.1109/ICIEV.2016.7760051

CITATIONS READS

11 4,978

4 authors, including:

Md Ashiqur Rahman Laskar Md. Shah Alam

Arizona State University University of Toledo
31 PUBLICATIONS 411 CITATIONS 7 PUBLICATIONS 87 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Md Ashiqur Rahman Laskar on 29 December 2017.

The user has requested enhancement of the downloaded file.

 5th International Conference on Informatics, Electronics and Vision (ICIEV), 2016 Publisher: IEEE DOI: 10.1109/ICIEV.2016.7760051

MODELING AND SIMULATION OF

LONGITUDINAL AUTOPILOT FOR
GENERAL AVIATION AIRCRAFT

Md. Tanjemul Islam1, Md. Shah Alam2, Md. Ashiqur Rahman Laskar3, Atul Garg4
1-4Department of Aeronautical Engineering

Military Institute of Science and Technology (MIST), Dhaka, Bangladesh

Email: [email protected], [email protected], [email protected], [email protected]

model a longitudinal autopilot for general aviation

Abstract- From the dawn of advanced aerospace technology,
aircraft. Autopilot is now the object of great interest in the
the aspect of stability and automated control is an important
field of research interest providing ease of handling and field of research because of its newly increasing
friendly assistance for piloting. Considering this challenging applications in aircrafts, missiles, space crafts, ships,
aspect, development of automated control system is chosen for UAVs and others. Previously various researches have
this research. The aim of this paper is to design a longitudinal been carried out on autopilot and the performances of
autopilot for a general aviation aircraft and to analyze the autopilot. One of the research on missile autopilot
performance of the developed model both for normal
describes about the design of longitudinal autopilot using
condition and atmospheric disturbance. To improve the
overall performance, stability of the system and to overcome back stepping approach. Two autopilot topologies are
the atmospheric disturbance a compensator along with a PID proposed in this approach one is angle of attack (AOA)
controller has been designed. Simulation results of MATLAB autopilot and another one is acceleration autopilot. The
Simulink have been also analyzed. nonlinear missile longitudinal dynamics is dealt with
Keywords- Elevator, Pitch controller, Autopilot, General
firstly to meet the strict feedback loop [2]. Another
aviation aircraft, MATLAB research has been done on the three loop autopilot design
and simulation. In order to realize missile real-time
change flight trajectory, a three loop autopilot has been
I. INTRODUCTION
set up. After the analysis of performance, the results show
An autopilot is designed depending on the control that autopilot gains played a good job in the flight
surfaces of the aircraft. They are broadly classified as trajectory, autopilot satisfied the flight index [3]. One
longitudinal autopilot and lateral autopilot. Pitch angle research shows linear active disturbance rejection control
and altitude of the aircraft is controlled by longitudinal approach for near space vehicle autopilot because the
autopilot. On the other hand, roll and yaw movements are conventional PID controller parameters are difficult to be
controlled through lateral autopilot. In case of lateral- decided on the dynamics of the near space vehicle within
directional mode, more than one control surface a large flight envelope. The approach is employed to a
contributes to perform the action. But in case of hypersonic near space vehicle autopilot. The numeral
longitudinal modes pitch angle and altitude depend only simulations show that the proposed approach has a good
on the contribution of elevator. For these reasons lateral performance on frequency domain and time domain. The
autopilot is much more complicated than the longitudinal angle of attack command is tracked very well, and the
autopilot. That’s why model of a longitudinal autopilot controller parameters are easy to be determined [4]. One
has been chosen in this paper which will control the of research has been done on design and simulation of
aircraft at normal condition and also in atmospheric longitudinal autopilot modes for a conventional aircraft.
disturbance by controlling pitch angle and altitude at the There are different modes of automatic longitudinal flight
same time. General aviation covers a large range of controls. Among those pitch attitude holds, altitude hold
activities, both commercial and non-commercial, and vertical speed hold mode autopilots are implemented
including flying clubs, flight training, agricultural in which controller is designed for each of these modes.
aviation, light aircraft manufacturing and maintenance By the use of this controller, the system performance is
[1]. Considering these factors, decision has been made to improved [5].
 5th International Conference on Informatics, Electronics and Vision (ICIEV), 2016 Publisher: IEEE DOI: 10.1109/ICIEV.2016.7760051

From the above studies it can easily be understood that mVT c c

autopilot is such a system which is very important to ( s − Cxu ) u + (− Cxα̇ s − Cxα ) α + (− Cxq s −
Sq 2VT 2VT
control air vehicles automatically and to reduce the work
load of human being. Considering these importance, an Cw (cos Θ)) θ = 0 (4)
attempt has been made in this paper to model such an
autopilot which can control the pitch angle and altitude of mVT c mVT
a general aviation aircraft. The designed autopilot of this −(Czu u) + [( − Czα̇ ) s − Czα ] α + [(− −
Sq 2VT Sq
paper is also able to compensate the sudden disturbance c
Czq ) s − Cw (sin Θ)] θ = 0 (5)
errors. Performances are analyzed for type 0 system in this 2VT

paper. If the aircraft displaced from its straight and level Iy

c c
flight condition by the strike of sudden gust, then how the −(Cmu u) + ( Cmα̇ − Cmα ) α + ( s2 − Cmq s) θ = 0
2VT Sqc 2VT
performance of the aircraft will change to keep the aircraft (6)
in its equilibrium is also analyzed here. Various
simulations have been done using MATLAB Simulink.
A quadratic equation is obtained after solving equation 4,
The results for the pitch angle response of the aircraft are
5and 6. The solution of this quadratic equation gives two
analyzed and the performance is measured in terms of rise
quadratic factors, one is short period mode and another
time (Tr), settling time (Ts), percentage overshoot (%OS), one is phugoid mode. The frequency of the oscillation for
steady state error (ess) for pitch angle response of the the phugoid is much lower than that observed for the
aircraft. short period mode. If the short period mode is unstable, it
II. AIRCRAFT DYNAMICS will generally be impossible for the pilot to safely control
the aircraft for any period of time. The period is so short
The xz-plane has been considered as a plane of symmetry. The that the speed does not have time to change, so the
performance of the aircraft can adequately be described by oscillation is essentially an angle- of-attack variation.
assuming the aircraft to a point mass concentrated at the Since this mode is usually highly damped and disturbance
aircraft’s center of gravity (cg). Then assuming that the origin will be considered in this research, its frequency and
of axis system is the center of gravity of the aircraft and damping are very important in the assessment of aircraft
perturbations from equilibrium are small. Also assuming that handling. Moreover, the short period frequency is
the earth and atmosphere are fixed in the inertial space and strongly related to the airplane's static margin. In the
mass (m) is constant (dm/dt = 0). The aircraft has been simple case of straight line motion, the frequency is
considered as a rigid body. Considering these, the longitudinal proportional to the square root of Cmα / CL. Therefore,
equations of motion can be written as following- only short period mode is considered here. Now
mVT c c
substituting u = 0 in the longitudinal equation of motion
( u̇ − Cxu u) + (− Cxα̇ α̇ − Cxα α) + (− Cxq θ̇ − since the short period oscillation occurs at almost
Sq 2VT 2VT
Cw (cos Θ) θ) = CFxa (1) constant forward speed. The equation along X axis does
not contribute to the short period oscillation. With these
assumptions and neglecting Cz α ̇ the equations become :
mVT c mVT
−(Czu u) + ( − Czα̇ ) α̇ − Czα α + (− −
Sq 2VT Sq
c
Czq ) θ̇ − Cw (sin Θ)θ = CFza (2) (
mVT
s − Czα ) α + [(−
mVT
−
c
Czq ) s − Cw (sin Θ)] θ =
2VT
Sq Sq 2VT

c Iy c
Cz ∂e ∂e (7)
(Cmu u) + ( Cmα̇ α̇ − Cmα α) + ( θ̈ − Cmq θ̇) = Cma c Iy c
2VT Sqc 2VT
(3) ( Cmα̇ s − Cmα ) α + ( s2 − Cmq ) θ = Cm ∂e ∂e (8)
2VT Sqc 2VT

Where u̇ represents the forward velocity, α̇ represents the

angle of attack and θ̇ represents the pitch angle. m, c, q, S Where ∂e is the deflection of elevator and Cz ∂e =
and VT represents mass, mean aerodynamic chord, dynamic (c/lt) Cm ∂e . From the geometrical figures of the aircraft
pressure, span area and total forward velocity in the the value of c/lt can be found [7].
longitudinal direction. Cxu, Cxα̇ , Cxα, Cxq, Cw, Czu, Czα ̇, Czα,
Czq, Cmu, Cmα ,̇ Cmα and Cmq are the stability derivatives [6].
To obtain the transient solution, consider the external
inputs CFxa= CFza= Cma= 0. Now, applying the Laplace
transform to equation 1-3 with initial condition zero yields-
 5th International Conference on Informatics, Electronics and Vision (ICIEV), 2016 Publisher: IEEE DOI: 10.1109/ICIEV.2016.7760051

between the PID controller output and input. Thus, the plant
TABLE I. VALUES OF THE COEFFICIENTS [7]
includes all blocks in the control loop, other than the
VT Cmq Cmδe Czα s Iy m controller itself. By just clicking “Tune” the system
automatically brings the PID Tuner tool. The resulting PID
176 -9.96 -.923 -4.49 184 3000 85.403 controller can be shown as:
C Czδe q Cmά Czq Cmα CwSinθ
1.191𝑒04𝑠 2 + 3.251𝑒05𝑠 + 2.218𝑒06
PID = (13)
5.7 -.189 36.82 -4.36 0 -.673 0 𝑠 2 + 1.669𝑒04𝑠

After substituting the values from the table, equations 7 and Simulink Modeling:
8 become: The reference pitch angle set for the autopilot is always
compared with the actual pitch angle at that moment. Thus
(2.2189 ρ + 4.49)α – (2.2189 ρ)θ =-0.1887 ∂e (9) error signal is produced which is fed to the controller and a
(-0.0706 ρ + 0.673)α + (0.07769 ρ +0.1613 ρ) θ = -0.923 ∂e
2 calculated correction is given by the elevator servo. Thus
desired pitch angle is maintained. For modeling the “Pitch
(10) displacement autopilot” MATLAB Simulink has been used.
Now, solving equation 9 and 10 the desired transfer function is The model of the Pitch displacement autopilot has been
obtained as following: developed as following-

θ(S) 11.8(S + 1.97)

=− 2
δe (S + 5S + 12.96)
(11)

III. MODELING PITCH DISPLACEMENT

AUTOPILOT

Compensator Design:
Compensators are not only used to improve the transient
response of a system but they are also used independently to
improve the steady-state error characteristics. To design the
compensator, automated tuning method is used using
MATLAB SISO Tool. The SISO Design Tool simplifies the
task of designing and tuning compensators. There are five Figure 1. Pitch displacement autopilot.
automated tuning methods in the SISO Design Tool to help
designing an initial stabilizing compensator for a SISO loop
Adding any kind of external gust or disturbance it can be
on-the-fly or refine existing compensator design so that it
further modified as,
satisfies a certain user-defined design specification. The
obtained compensator’s transfer function is,

49.394*((1+.08*s)*(1+.39*s+(.28*s)^2))
C = ………………………………………...... (12)
s*(1+.51*s)*(1+.035*s)

PID Controller Design:

Controller is the element in the system which regulates,
directs or commands the entire plant according to the error
signal generated. The main task of a controller is to monitor
the dynamic system behavior and direct the system to the
required operating condition. The PID Tuner automatically
computes a linear model of the plant in the Simulink model.
It considers the plant to be the combination of all blocks Figure 2. Pitch displacement autopilot with disturbance.
 5th International Conference on Informatics, Electronics and Vision (ICIEV), 2016 Publisher: IEEE DOI: 10.1109/ICIEV.2016.7760051

In this final Simulink model of pitch displacement or

longitudinal autopilot, reference is a constant block where
autopilot command will be put. Inner closed loop has
simply a unity feedback gain. The change in pitch attitude
due to atmospheric disturbance like gust is sensed by a
gyroscope and the gyroscopic detected signal is fed at the
input of inner loop. The Inner loop provides the support to
reduce errors at a minimum that occur in control surface
actuator or aircraft dynamics since those are not ideal.
Outer closed loop gives the final correction to maintain
aircraft pitch attitude (angle) according to the autopilot
Figure 4. Response for Pitch up (+10 deg.) command condition without
command. To see the final response of the whole system a disturbance.
scope block has been added at the end of the model.

IV. SIMULATION RESULT ANALYSIS

Simulation is the imitation of the operation of a real-

world process or system over time [8]. In this study, all
the simulations are done for 10 sec.
Response in different pitch angle command conditions
without disturbance:
At the early portion of this result analysis, considering
that a general aviation aircraft is flying where there is no Figure 5. Response for Pitch down (-3 deg.) command condition without
disturbance like gust. Now, three types of command disturbance.
condition have been considered for pitch control
autopilot. Autopilot pitch command (Reference) as 0
From the above figures, it is observed that rise time is 14.969
degree for level flight, +10 degree for pitch up condition,
ms and percentage of overshoot is 0.069% for pitch up
-3 degree for pitch down condition. The response of the
condition. For pitch down condition rise time (Fall time) is
designed model for each condition is simulated and
14.968 ms and percentage of overshoot is 1.994 %. So,
output theta, ɵ (Pitch angle) is taken from the scope block
designed model is more reliable in pitch up condition than
which are shown below respectively. Simulating the
pitch down condition. Also, level flight response is steady and
model shown in fig-1, following three figures are
perfect.
obtained.

Response in different pitch angle command conditions with

standard disturbance:
In this part comparison has been made between conventional
model and designed model by considering the responses at
different pitch angle command conditions. Here disturbance is
fed at the input of the inner loop. Unit step is used as standard
input which is acting as the gyroscope detection of aircraft
pitch angle change due to gusts or disturbances. Simulating the
models shown in fig. 1 & fig. 2, following six figures are
obtained.

Figure 3. Response for level flight (0 deg.) command condition without

disturbance.
 5th International Conference on Informatics, Electronics and Vision (ICIEV), 2016 Publisher: IEEE DOI: 10.1109/ICIEV.2016.7760051

Figure 10. Response of the designed model at pitch down command condition
(Pitch angle= -3 deg.).
Figure 6. Response of the designed model at level flight command
condition (Pitch angle=0 deg.).

Figure 11. Response of the conventional model at pitch down command

Figure 7. Response of the conventional model at level flight command condition (pitch angle=-3 deg.).
condition (Pitch angle=0 deg.).

The above Simulation graphs clearly represent that newly

designed automatic pitch control model has advantages over
previous one with highly reduced percentage of overshoot and
rise time. For example, at +10 degree pitch up condition some
parameters are listed below:

TABLE II. PARAMETERS OF PERFORMANCE CHARACTERISTICS

Parameter Designed Conventional

model model
Rise Time 15.120 ms 198.862 ms

Figure 8. Response of the designed model at pitch up command condition

% Overshoot 0.312 % 32.667%
(Pitch angle= +10 deg.).
% Preshoot 0.505% 0.667%

% Undershoot 1.553% 10.501%

Response in different pitch angle command conditions

with rectangular pulsed disturbance:

Now rectangular shaped pulse is introduced instead of

standard input as disturbance. Pulse width of it is 3 sec and
amplitude is 1. Pulse width indicates the duration of faced
disturbance.
Figure 9. Response of the conventional model at pitch up command condition
(Pitch angle= +10 deg.)
 5th International Conference on Informatics, Electronics and Vision (ICIEV), 2016 Publisher: IEEE DOI: 10.1109/ICIEV.2016.7760051

On the contrary, conventional model requires 4.4 sec (3-7.4

sec) to compensate with 3 sec duration disturbance or gust. As
a result, designed pitch control or longitudinal autopilot can
provide piloting the general aviation aircraft more effectively.

V. CONCLUSION

PID controller Tuning method is utilized to design the PID

controller where maximum faster time response and robust
transient behavior are emphasized. This designed PID
controller has dynamic response compared to other normal
Figure 12. Rectangular shaped pulse as disturbance. controllers. Due to the limitations of the open loop system, a
closed loop (Outer loop) system is introduced with efficient
Simulation is repeated and response is observed. This time characteristics using PID controller. Pitch angle position
only level flight that means 0 degree pitch angle command feedback is used in the outer loop of pitch control to reduce
condition is considered. Designed model shows more stable the effect of atmospheric disturbances like gusts. The stability
response than conventional model. Amplitude of response of of the whole systems is checked and found as stable. Then
designed model is 0.04 while 0.17 for other. From the Simulink model is generated and simulated for three different
following figures, it can be seen that designed model takes autopilot command conditions. The longitudinal autopilot is
total 1 sec to compensate a disturbance (Gust) of 3 sec also tested in the presence of atmospheric disturbances.
duration where 0.5 sec at the beginning(3-3.5 sec) of gust and Comparison between conventional and designed model
rest 0.5 sec at the ending(6-6.5 sec) of gust. represents that designed model requires only 1 second for
compensating a 3 seconds duration gust where conventional
model takes 4.4 seconds. The designed longitudinal autopilot
possesses both rise time and settling time in milliseconds with
less than 0.5% overshoot. Finally, it can be concluded that a
tremendous improvement in overall performance and
robustness have been achieved to withstand the desired
conditions.

REFERENCES

[1] Crane, Dale (1997). Dictionary of Aeronautical Terms, third edition,

page 238-239. Aviation Supplies & Academics,. ISBN 1- 56027-287-2.
[2] Fan Jun-fang, Su Zhong, “Missile longitudinal autopilot design using
Figure 13. Response of the designed model at level flight backstepping approach”, Aerospace Conference, Big Sky, MT , 6-13
command condition (Pitch angle=0 deg.). March 2010.
[3] Chu Hai-rong, Zhang Yue, “Three-loop autopilot design and
simulation”, International Conference on Mechatronics and Automation
(ICMA), Chengdu, 5-8 Aug. 2012.
[4] Hui Li, Zhide Yin, Yu Fan, Fengyi Li, Chuang Song, “A linear active
disturbance rejection control approach for near spacevehicle autopilot”,
International Conference on Control, Automation and Information
Sciences (ICCAIS), Changshu, 29-31 Oct. 2015.
[5] Solomon Raj, K.D.; Kumar, P.R., “2nd International Conference on
Electronics and Communication Systems (ICECS) ”, Coimbatore, 26-27
Feb. 2015.
[6] S.Goyal, “Study of a Longitudinal Autopilot For Different Aircrafts”,
B.E project report, Punjab Engineering College (DU) Chandigarh,
INDIA, 2006.
[7] R.C. Nelson, Flight Stability and Automatic Control, McGraw-Hill,
1997.
Figure 14. Response of the conventional model at level flight command
condition (Pitch angle=0 deg.). [8] J. Banks, J. Carson, B. Nelson, D. Nicol (2001). Discrete-Event System
Simulation. Prentice Hall. p.3.ISBN 0-13-088702-1.

View publication stats