Test Platform to Pitch Angle using Hardware in Loop

advertisement
Session T1A
Test Platform to Pitch Angle using Hardware in
Loop
Sérgio Ronaldo B. Santos and Neusa Maria F. Oliveira.
Instituto Tecnológico de Aeronáutica, sronaldo@ita.br, neusa@ita.br.
Abstract - The low cost electronic component and sensors
makes the development of UAVs more accessible.
Consequently, more students are interested in the skills
necessary to integrate an UAV developing group. A very
important component in a UAV is the automatic pilot.
Aiming Education and Research in Digital Automatic
Pilot development a flexible system to test flight
controllers is proposed. The system uses the structure
known as “hardware in the loop” in which the aircraft
model is simulated in a PC computer and the designed
controller is implemented in a microcontroller. The
proposed system is appropriate to be used in laboratory
classes in which the students can test the controllers
previously designed using control theory and can also
practice the use of interfacing devices, necessary to
convert the analog signals provided by the sensors to the
digital form used by the processors, and communication
devices, necessary to the information exchange between
the PC and the microcontroller. The laboratory practice
is such that the student can compare the results obtained
using the conventional simulation approach, in which a
control loop consisted of the designed controller and the
aircraft model transfer function is simulated in a
computer, and the results obtained using the proposed
platform. Implementing the designed controller in a
microcontroller that can be embedded in an UAV
exposes the student to a more real situation, considering
also the possible delays between a command given by the
controller and the change of a parameter in the aircraft.
The results presented here were obtained implementing a
PID controller in a microcontroller and doing all the
connections necessary to a PC implementing a
longitudinal motion model of an aircraft.
new version.
Index Terms - Pitch displacement autopilot, Pitch angle, PID
control, Longitudinal autopilot.
NOMENCLATURE
θref
e
δc
δe
θ
Kp
Kd
Ki
=
=
=
=
=
=
=
=
Pitch angle reference
Erro signal
Output PID control
Elevator deflection angle
Airplane pitch attitude angle
Proportional control gain
Derivative control gain
Integral control gain
/09/$25.00 ©2009 IEEE
v
τs
= Elevator servo voltage
= Servo time constant
INTRODUCTION
Modern aircraft have many control systems that help pilots
during the flight and also the aircraft stability characteristics.
An automatic pilot is a device that controls an aircraft in a
pre-established flight without human intervention and has
the main function stabilizing the dynamic characteristics of
the aircraft. These tasks are performed by the control of
command surfaces [1].
An automatic pilot is also present in UAV (Unmanned
Aircraft Vehicles). Nowadays, UAV’s are in constant study
by several research groups in the world. Consequently, the
interest in studying automatic pilot has also increased.
Aiming the study of automatic pilot in laboratory, a platform
has been proposed and developed to test controllers designed
by students. The platform uses the structure known as
“hardware in the loop” and permits the designed controller
to be implemented in a processor that theoretically could be
embedded in the aircraft [2].
The proposed and developed test platform of aircraft
controllers is introduced herewith. Firstly, the problem that
is being addressed - the structure of the loop to be
implemented in the “hardware in the loop” - is exposed.
Then, we expose how the problem was solved, the tools used
and how to connect them to work together in order to
implement aircraft control loops. Finally, we show the
results obtained using the platform with a PID controller to
control the pitch angle of a generic aircraft [3].
PROBLEM STATEMENT
The development of a controller involves the knowledge of
control theory and simulations in a computer before, if
possible, the implementation in a physical systeme.
A test platform for controllers is introduced herewith. It
shall give an opportunity of implementing the designed
controller in a microcontroller that hypothetically should be
embedded in an aircraft. The proposed platform has the
structure hardware in the loop. This structure is such that the
aircraft model to be controlled is implemented in a PC
(personal computer) and the designed controller is
implemented in another processor [4].
The communication between the two blocks must be
done so that the system works properly. The communication
is done previously thus letting the student deal exclusively
with the controller implementation.
October 18 - 21, 2009, San Antonio, TX
39th ASEE/IEEE Frontiers in Education Conference
T1A-1
Session T1A
In the laboratory, a PC, a microcontroller, and
connecting cables are readily available to each student. The
longitudinal and lateral models of the aircraft with which it
is desired to work should be already available in the PC.
With this structure, the task to be developed in the laboratory
is the implementation of the designed controller in a
microcontroller that should be embedded in an aircraft. The
necessary interfacing between the PC and microcontroller
should be done by the student if computer interfacing is
defined as one goal of the laboratory class.
The study of aircraft control in general classifies the
movements realized by the vehicle into two instances: lateral
and longitudinal. In certain flight conditions it is not possible
to study these movements separately; however, in this study
it will be assumed flight conditions that permit such
consideration [5].
Figure 1 represents a general autopilot control loop and
shows the tasks done by PC and the microcontroller in the
test platform. The comparator and the digital controller are
implemented in the microcontroller. Depending on the
aircraft and the movement being studied, a different aircraft
transfer function and servo transfer functions are
implemented.
the PSoC, the modules in the CY8C29466 that were used in
the developed system are: UART, comparator, timers,
PWM´s, AD, and DA converters, 2Kbytes RAM data
memory and 16Kbytes program flash memory [8].
Considering Figure 1, which shows the structure of the
proposed platform, we present all the connections that were
made. One input signal to the microcontroller is the
reference signal yref, which should be followed by the output
signal y. yref is an analog DC signal between 0V and 5V and,
since it is analog, it is necessary to use an AD converter in
order to be handled by the microcontroller. The system
output signal is also a signal input in the microcontroller. It
compares the reference and output system signals and
generates the error signal to be used by the designed
controller to calculate the control signal to be sent to the
aircraft. In Figure 2 and Figure 3 the block diagrams
implemented in MATLAB/SIMULINK.
FIGURE 2
DYNAMIC FOR THE GENERIC AIRCRAFT PITCH CONTROL SYSTEM.
FIGURE 1
THE CONTROL LOOP TO BE IMPLEMENTED IN THE PROPOSED TEST
PLATFORM.
To study each one of these movements it is necessary to
specify the aircraft model of the aircraft in study. The design
of a controller for an aircraft is developed in order to control
one or more flight parameters such as altitude, speed, etc [7].
In an aircraft, it is possible to have several control loops. In
the development of this test platform, only the lateral and
longitudinal loops are taken into consideration, and the
controllers are designed for each of them depending on the
characteristics to be controlled. The controller should be
designed before the laboratory class takes place. The
problem resolution and the results presented consider the
longitudinal movement of a generic aircraft.
FIGURE 3
SERIAL MODULE COMMUNICATION.
The S-Function block was used to establish the serial
communication between the aircraft model and the PID
controller. All the parameters necessary for the
communication were configured in a scrip file (m-file) and
this file is virtually connected to the S-Function. Another
important parameter defined in this file was the sample time
of the aircraft output signal. It was adjusted to 500ms.
Since the output signal is able to be sent back to the
microcontroller, the control loop is closed and the platform
is ready to work.
IMPLEMENTATION AND RESULTS
The control test platform was developed and is ready to be
used in laboratory. The example chosen to be presented here
PROBLEM RESOLUTION
was a control system for the longitudinal movement of an
In this section we discuss the development of the proposed aircraft. A pitch control loop for a generic airplane [7] was
platform. The microcontroller chosen is the PSoC implemented in the developed platform and the results are
(Programmable System on Chip). Several interfacing commented here.
As it has already been mentioned, the proposed system
devices are integrated within this chip thus
minimizing
is
to
be used in laboratory to test designed controllers.
the need of external components and the space required in
the PCB (Printed Circuit Board). Among others present in Therefore, the student should study control theory, design
the controller and test it using computer simulation. Having
/09/$25.00 ©2009 IEEE
October 18 - 21, 2009, San Antonio, TX
39th ASEE/IEEE Frontiers in Education Conference
T1A-2
Session T1A
finished the process, it will be implemented in the proposed
platform in the laboratory. Consequently, the control theory
issues are not the main objective on this paper and they will
be presented as briefly as possible.
Consider a closed-loop control system as in Figure 1,
representing the pitch control loop for a generic airplane [7].
The aircraft transfer function to be used is the transfer
function from the elevator deflection to the pitch angle, as in
(1).
(5)
Designing a PID using a continuous transfer function,
results in a continuous controller. In order to implement the
continuous PID controller on a digital device it is necessary
to convert it into a digital PID controller. One approach used
to obtain a digital controller with the same behavior as the
continuous controller is implementing in the digital
processor the equations (6) and (7).
(1)
(6)
The actuator involved in a pitch control loop is the
elevator servo. The elevator servo considered has a first
order transfer function, as in (2) and (3).
(2)
(3)
The relationship between the control signal, δc(s),
generated by the controller and the error signal, e(s), is
denominated the control law. The control law is used by a
PID controller. It executes three control actions,
proportional, integral and derivative actions, and can be
written as in (4) and (5).
(7)
Where T is the sample time used in the system. The
control test platform implementing the chosen pitch control
loop can be represented, as in Figure 4.
The results presented here were obtained applying a step
signal in the reference input θref. The behaviors of the system
output, θ(t), versus Time (s) are shown in graphs. The
maximum value that either, the input signals and output
signals, can reach is 5V and in the graphs it is equivalent to
1 (or 100% of the maximum). To this maximum input value
the deflection obtained in the elevator servo is 60o.
Figure 5 shows the response to an input step of
magnitude 1 in θref. The system presents an overshoot and
the pitch angle of the aircraft reaches the input value, θref,
after 39 sec.
(4)
FIGURE 4
BLOCK DIAGRAM OF THE PLATFORM IMPLEMENTING THE PITCH CONTROL LOOP
/09/$25.00 ©2009 IEEE
October 18 - 21, 2009, San Antonio, TX
39th ASEE/IEEE Frontiers in Education Conference
T1A-3
Session T1A
FIGURA 5
RESPONSE TO A STEP INPUT OF MAGNITUDE 1 FOR THE PITCH AUTOPILOT .
Figure 6 shows the response to an input step of
magnitude -0.4 in θref. The system presents an overshoot and
the pitch angle of the aircraft reaches the input value, θref,
after 26 sec, resulting in an elevator deflection angle final of
-24o.
As it can be seen in Figure 5, Figure 6 and Figure 7, the
output signal, θ(t), reaches the reference after some time.
Running the system, the observation of the movement of the
servo motor can be done and its final position compared
with the expected position related to the desired pitch angle.
Even though it is not presented here, the comparison of
the responses obtained by the control loop being simulated
all together in MATLAB and in the platform was done. It
was verified that the settling time of the output signal is
bigger when the system is implemented in the platform than
when it is simulated altogether in the MATLAB. It happens
because the discrete time process which takes place when
using the platform the PID implemented in the
microcontroller uses sampled values obtained in periods preestablished by the AD converter and the communication
block in SIMULINK. Due to these facts, the error signal
lasts more to reach the desired zero value. However, it is
important to notice that the problem which happens as a
consequence of the data sampling will be always present if
the problem is to control a continuous system with a digital
device. Therefore, this is another aspect with which the test
platform contributes in the learning process - the observation
of data samplings consequences in dynamic systems.
CONCLUSION
FIGURA 6
RESPONSE TO A STEP INPUT OF MAGNITUDE -0.4 FOR THE PITCH AUTOPILOT
.
Figure 7 shows the aircraft response to an initial step of
magnitude 0.7 in θref, resulting in an elevator deflection
angle of 42o. A second step was given in θref, this one with
magnitude -0.5, which results in an elevator deflection angle
of -30o.
The test platform for aircraft controllers was developed and
the results obtained from the implementation of a pitch
control loop of a generic aircraft were presented. The
platform is ready to be used in laboratory classes to improve
the study of automatic pilots, providing the experience of
implementing a controller in a device that could be the
automatic pilot computer. With the test platform it is
possible to practice the implementation of a controller in a
digital device, to experience the effects of discretizing the
signal and also realizes the visual observation of a servo
motor positioning movement – which to be performed, it is
necessary that the student controls the servo motor by PWM
signal.
The control loops for the longitudinal movements of an
aircraft can be implemented using the platform and the
hands on experience shall enhance the learning process.
REFERENCES
FIGURA 7
RESPONSE TO A INITIAL STEP INPUT OF MAGNITUDE 0.7 AND TO A INPUT
STEP OF MAGNITUDE -0.5 AT TIME T=65S.
/09/$25.00 ©2009 IEEE
[1] Bryson, A E, “Control of Spacecraft and Aircraft”. Princeton Univerty
Press, 1994.
[2] K.P. Valavanis,“ Introduction in Unmanned Aerial Vehicles: State of the
Art and the Road to Autonomy ”, Vol. 33, Intelligent Systems,Control, and
Automation: Science and Engineering, K.P. Valavanis, Ed.Springer: 2007,
pp. 23-34.
[3] P. Fabiani, V. Fuertes, A. Piquereau, R. Mampey and F. Konigsbuch,
“Autonomous flight and navigation of VTOL UAVs: from autonomy
demonstrations to out-of-sight flights” in Aerospace Science and
Technology, Vol. 11, Issues 2-3, p:183-193, March-April 2007.
[4] A. Ryan et al., “ An overview of emerging results in cooperative UAV
control. Decision and Control”,2004. CDC. 43rd IEEE Conference on,
2004, pp. 602-607 Vol.1.
October 18 - 21, 2009, San Antonio, TX
39th ASEE/IEEE Frontiers in Education Conference
T1A-4
Session T1A
[5] D. Jung e P. Tsiotras, “Modeling and Hardware-in-the-Loop Simulation
for a Small Unmanned Aerial Vehicle”, AIAA Infotech at
Aerospace,Rohnert Park, CA: 2007.
[6] R. Gimenes et al., “Using Flight Simulation Environments with AgentAutonomous
Robot
Systems
and
Controlled
UAVs”,
Competitions:Proceedings of the 8th Conference, Aveiro, Portugal: 2008.
[7] Nelson, R. C.,” Flight Stability and Automatic Control.”, McGraw-Hill
Second Editions, 1998.
[8] Cypress Semiconductor, “Technical Reference Manual (TRM) PSoC
CY8C29x66”, Version 1.0, 2006.
[9] Ogata, K, “Engenharia de Controle Moderno.” Prentice Hall do Brasil
Quarta Edição, 2003
[10] Hermely M.E., “Controle por Computador de Sistemas Dinâmicos”,
Edgard Blucher, 1996.
[11] Pallet, E. H. J., “ Automatic Flight Control ”, Blackwell Science Ltda,
1993.
[12] Roskam, J., “Airplane Flight Dynamics and Automatic Flight
Dynamics and Automatic Flight Controls, PART I”., Roskam Aviation and
Engineering Corporation, 1982.
[13] Rauw, M. “FDC 1..2 A Simulink Toolbox for Flight Dynamics and
Control Analysis”, 1999.
[14] Lobo, J. F., “ Noções de Física Aplicada a Vôo Simulado.”, Fórum de
Simulações Aéreas ( FSA), http:// www.fsa.com.br .
[16] Blakelock, J. H., “Automatic Control of Aircrafts and Missiles”,
Wiley,1991.
/09/$25.00 ©2009 IEEE
October 18 - 21, 2009, San Antonio, TX
39th ASEE/IEEE Frontiers in Education Conference
T1A-5
Download