Professional Documents
Culture Documents
Modeling and Controller Design of A VTOL UAV: June 2015
Modeling and Controller Design of A VTOL UAV: June 2015
net/publication/281461479
CITATIONS READS
19 4,699
8 authors, including:
Some of the authors of this publication are also working on these related projects:
Development of a Thrust Vectoring System with Coanda Effect / Coanda Etkisi ile Çalışan İtki Yönlendirme Sistemi Geliştirilmesi View project
All content following this page was uploaded by Dilek Funda Kurtulus on 09 September 2016.
A. Propulsion Model
The propulsion system model is based on a classical blade
element/vortex formulation. An open source code, QPROP
[5], is used to obtain the torque-thrust properties of the
propellers employed. For this, the geometry of the propellers,
the spin rate as well as the axial velocity the propellers face,
are used as inputs.
To verify the QPROP code and determine the 5 m/s wind speed
performance characteristics of the electric motor-propeller
combinations, wind tunnel tests are also performed in the (1m
× 1m) test section wind tunnel of the Aerospace Engineering
Department for different rotor spin axis angles (0○, 20○, 40○,
60○) and wind speeds (0, 5, 8, 10, 12, 15, 20, 30 m/s). The
wind tunnel and the test setup are shown in and Figure 3. and
Figure 4. respectively. Wind tunnel corrections are applied to
the velocity results obtained [6].
8 m/s and 12 m/s wind with 40o and 60o main rotor tilt
angles, since these corresponds to stall speeds mainly
encountered in the transition flight phase. A database is
created using the measurements of the thrust and torque
values at different RPMs, rotor angles and wind speeds. A 10 m/s wind speed
data acquisition system is used to record the load cell data Figure 5. Comparison of Thrust in QPROP with Experimental Data for 0○
degree tilt angle
10 m/s wind speed
5 m/s wind speed
Parameter Inputs
Control Surface -20o < δaileron <20o
(deg) ∆β =30o
Parameter Inputs
Control Surface -20o < δelev <20o TABLE V. PITCHING MOMENT COEFFICIENT (CM) OF THE TAIL FOR
Deflections ∆δelev =10o DIFFERENT ANGLES OF ATTACK AND ELEVATOR DEFLECTIONS
(elevator)
Control Surface -20o < δrudder<20o
Deflections (rudder) ∆δrudder=10o Control Surface Deflection (deg)
Angle of attack -16o <α <16o
(deg) ∆α =2o
Beta -60o < β <60o
(deg) ∆β =30o
Angle of attack (deg)
𝑝̇
Angle of attack (deg)
𝑝 0
𝑞̇ 𝑞 0
𝑟̇ 𝑟 0 (3)
[𝐴 [𝐵𝐾] 𝜙
𝜙̇ = − 𝐵𝐾] 𝜙 + 𝑐𝑜𝑚
𝜃̇ 𝜃 𝜃𝑐𝑜𝑚
[𝜑̇ ] [ 𝜑] [𝜑𝑐𝑜𝑚 ]
Phi (deg)
20 Roll Command
[𝜑̇ ] [𝜑]
10
Where, 𝑝, 𝑞, 𝑟 are the body angular rates, 𝜙, 𝜃, 𝜑 are the 0
Euler angles of the aircraft. For the tracking problem, a
-10
slightly different algebraic Riccati equation is used 0 5 10 15 20 25 30
Time (sec)
(Eqn.(9)).
Figure 11. Roll angle command and response without washout filters - LQR
(9)
𝐴𝑇 𝑆 + 𝑆𝐴 − 𝑆𝐵𝑅−1 𝐵𝑇 𝑆 + 𝐶 𝑇 𝑄𝐶 = 0
40
30
(10) Pitch Angle
Theta (deg)
−1 𝑇
𝐾=𝑅 𝐵 𝑆 20 Pitch Command
10
(11)
𝐾𝑧 = 𝑅−1 𝐵𝑇 [𝑆𝐵𝑅−1 𝐵𝑇 − 𝐵𝑇 ]−1 𝐶 𝑇 𝑄 0
-10
0 5 10 15 20 25 30
The gain matrices used are given below: Time (sec)
Figure 12. Pitch angle command and response without washout filters –
LQR
30
20
Psi (deg)
Yaw Angle
Yaw Command
10
0 5 10 15 20 25 30
Time (sec)
V. SIMULATION RESULTS
Figure 13. Yaw angle command and response without washout filters –
The designed linear controller is examined first through LQR
the simulation code. In order to observe the performance of
the controller, pilot commands and the air vehicles responses
in roll, pitch and yaw channels are given in this section. 40
30
The pilot commands are given as a pulse input of 30 Roll Angle
Phi (deg)
degree between 5-7 seconds time interval for roll, pitch and 20 Roll Command
yaw the Euler angles responses using linear quadratic 10
regulator controllers are presented in figures 11-13. 0
The filter is converted into discrete form using Tustin’s 20 Pitch Command
approximation [9] to implement in the microcontroller. The 10
results with washout filter are given in figures 14-16. 0
Comparing the results presented in the figures it may be
-10
concluded that in all cases, the response follows the 0 5 10 15
Time (sec)
20 25 30
20 Yaw Command
investigate the rolling, pitching and yawing motion of the air
10 vehicle. When the vertical load on the bar is eliminated, the
0 bar may slide up (retract or shorten) as the vehicle lifts off.
-10
0 5 10 15 20 25 30 To test the controller, a sufficient throttle is given to lift
Time (sec)
the aircraft slightly. Then, the roll, pitch and yaw commands
Figure 16. Yaw angle command and response with washout – LQR are applied by the remote controller. Both the Euler angle
command and Euler angle responses measured by the IMU
Simulations with LQT controller are also carried out. The on board, captured by the microcontroller, and transmitted to
results are presented in figures 17-19. From these figures it the laptop used as the ground station through the telemetry
may be observed that in all channels, the aircraft is quite system. The results are presented in figures 21-23. LQR
fast. Thus, both the rise times and the over shoots controller in used in these hover tests presented. These tests
carried out in the laboratory shows that the system is
encountered are much smaller than the above controllers. In
operating properly, and the commands are followed quite
fact with LQT, the system tracks the control inputs quite closely.
closely.
In the next section flight test results are presented.
40
30
Roll Angle
Phi (deg)
20 Roll Command
10
-10
0 5 10 15 20 25 30
Time (sec)
30
Theta (deg)
Pitch Angle
20 Pitch Command
40
10 Roll angle
Roll command
20
0
Phi (deg)
-10 0
0 5 10 15 20 25 30
Time (sec)
-20
Figure 18. Pitch angle command and response - LQT
-40
40 0 10 20 30 40 50
Time (sec)
30
Yaw Angle
Figure 21. Roll command and roll response of the aircraft - LQR
Psi (deg)
20 Yaw Command
10 40
Pitch angle
0 Pitch command
20
Theta (deg)
-10
0 5 10 15 20 25 30 0
Time (sec)
-40
VI. HOVER TEST RESULTS 0 5 10 15 20 25 30
Time (sec)
The hover tests are first carried out in the laboratory. The Figure 22. Pitch command and pitch response of the aircraft - LQR
test equipment is shown in Figure 20. , is composed of a
sliding bar attached to the ceiling of the test room at one end
and the center of gravity of the air vehicle at the other end.
Mechanism is simply working like a pendulum when the
work is to develop control algorithms for going from vertical
40 take-off to forward flight and landing again back vertically.
20
Psi (deg)
0
40
-20
Phi (deg)
20
-40 Yaw angle
Yaw command
0
-60
0 5 10 15 20 25
Time (sec) -20 Roll angle
Roll command
Figure 23. Yaw command and the yaw angle response of the aircraft - LQR 0 5 10 15 20 25 30 35 40 45
Time (sec)
Theta (deg)
commands are not obeyed by the aircraft and there was 20
difficulty in steering the aircraft. 10
noted that, during the last flight test, the weather was quite 0 10 20 30 40 50 60 70 80
Time (sec)
windy. Hence, the pilot was giving rapid commands to keep
the aircraft from drifting away, while in the yaw channel, the
pilot’s commands were much slower. Thus the controller did Figure 26. Commanded and realized yaw angles - LQR controller
not have time to respond to the commands, creating an offset
between the commands and the responses shown in the 20 Roll angle
figures. Roll command
10
Of these three controllers, LQT controller performed the
Phi (deg)
washout filter in the autopilot [11]. Pictures from flight tests -10
are given in Figure 33.
-20
0 10 20 30 40 50 60 70
VII. CONCLUSION Time (sec)
Figure 27. Commanded and realized roll angles - LQR controller with
In this study a nonlinear dynamic model is developed for washout filter
a tri-copter VTOL UAV. A linear quadratic regulators and
linear tracking controllers are designed to stabilize the
aircraft in the takeoff-hover-land flight phases. The success
of the designs are demonstrated through the nonlinear
simulations as well as the flight tests carried out.
A linear controller is being developed to control position
of the aircraft in the hover flight. The ultimate goal of this
20
Pitch angle
10 Pitch command
Theta (deg)
-10
-20
0 5 10 15 20 25 30
Time (sec)
Figure 28. Commanded and realized pitch angles - LQR controller with
washout filter
40
Yaw angle
Yaw command
20
Psi (deg)
-20
0 20 40 60 80 100
Time (sec)
Figure 29. Commanded and realized yaw angles - LQR controller with
washout filter
20
Roll angle
Roll command
10
Phi (deg)
-20
5 10 15 20 25 30 35 40 45 50
Time (sec) REFERENCES
[1] Armutcuoglu, O, Kavsaoglu, M.S. and O. Tekinalp, “Tilt Duct
Figure 30. Commanded and realized roll angles with LQT controller Vertical Takeoff and Landing Uninhabited Aerial Vehicle Concept
Design Study,” AIAA Journal of Aircraft, vol. 41, No. 2, pp. 215-223,
March-April 2004.
40 Pitch angle [2] Okan, A., Tekinalp, O., and M. Kavsaoglu, “Flight Control of a Tilt-
Pitch command
30
Duct UAV,” 1st International Conference on Unmanned Aerospace
Vehicles, AIAA-2002-3466, Portsmouth, Virginia, 20-23 May 2002.
Theta (deg)
20
[3] Şenipek M., Yayla M., Limon A.U., Rouzbar R., Yosheph Y., Kalkan
10 U., Şenol N., Akel E., Güngör O., Hoş B., Usta A., Uzunlar İ.Ö.,
0
Sarsılmaz S.B., Kurtuluş D.F.(2013), “Design Process of an UAV for
AIAA DBF 2013 Competition,” 7th Ankara International Aerospace
-10 Conference, Ankara, Turkey, 11-13 September 2013, Paper No.
0 10 20 30 40 50 AIAC-2013-105,.
Time (sec)
[4] Department of Defence World Geodetic System 1984. Available at:
Figure 31. Commanded and realized pitch angles with LQT controller http://earth-info.nga.mil/GandG/ publications/tr8350.2/ wgs84fin.pdf
[5] M. Drela and M. I. T. Aero & Astro, “QPROP Formulation” June,
2006. Avaliable at: http://web.mit.edu/drela/Public/web/qprop/.
[6] Barlow, J., Rae, Jr., W., & Pope, A. Low-Speed Wind Tunnel Testing
40 (3rd ed.). John Wiley & Sons, 1999.
30 [7] Önen A. S., Şenipek M., Mutlu T., Cevher L., Güngör O., Kurtuluş
Yaw angle
Yaw command Bozdemir D. F., Tekinalp O.. “The wind tunnel tests for the
20
propulsion system of a VTOL UAV”, HİTEK, Istanbul, Turkey, 18-19
Psi (deg)
10
June 2014.
0 [8] C. A. Deperrois, “About XFLR5 calculations and experimental
-10 measurements,” October, 2009, Available at:
-20 http://www.xflr5.com/docs/Results_vs_Prediction.pdf .
[9] D. G. Hull, Optimal Control Theory for Applications, Springer, 2003.
0 10 20 30 40 50 60 70
Time (sec) [10] Prach, A., Tekinalp, O., “Development of a State Dependent Riccati
Equation Based Trracking Flight Controller for an Unmanned
Figure 32. Commanded and realized yaw angles with LQT controller Aircraft,” AIAA Guidance, Navigation and Control Conference,
Boston, MA, 19-22 August 2013
[11] Önen, A.S., Modeling and Controller Design of a VTOL Air Vehicle,
M.S. Thesis, Middle East Technical University, Department of
Aerospace Engineering, Ankara, Turkey, March 2015.