You are on page 1of 30

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

net/publication/366177309

Mathematical Modeling and Pitch Control of 1-DOF tail plane

Technical Report · June 2021


DOI: 10.13140/RG.2.2.13274.62407

CITATIONS READS
0 9

9 authors, including:

Taha Mahmood Talha Bin Asad


Ghulam Ishaq Khan Institute of Engineering Sciences and Technology Ghulam Ishaq Khan Institute of Engineering Sciences and Technology
10 PUBLICATIONS   0 CITATIONS    1 PUBLICATION   0 CITATIONS   

SEE PROFILE SEE PROFILE

Ubaid Ullah Ume Aimen


Ghulam Ishaq Khan Institute of Engineering Sciences and Technology Ghulam Ishaq Khan Institute of Engineering Sciences and Technology
1 PUBLICATION   0 CITATIONS    1 PUBLICATION   0 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Surface tension of water using Lucas-Washburn equation View project

solar pond power plant View project

All content following this page was uploaded by Taha Mahmood on 10 December 2022.

The user has requested enhancement of the downloaded file.


ME-464 System Dynamic and Control
Complex Engineering Problem

Members:
Taha Mahmood 2017472
Talha Bin Asad 2017473
Ubaid Ullah 2017480
Ume Aimed 2017481
Wajid Ali Khan 2017498
Zabidullah 2017504
Zia Ullah 2017515
Saad Ahmed Toor 2017905
Saad Hassan Khera 2017906
Submitted to: Sir S.M Ahmed
Abstract
The present study deals with modeling and pitch control of one degree of freedom tail plane from
the experimental provided data and to create a feedback loop system using controller that
satisfied basic performance criteria such as stability, reference tracking, disturbance rejection,
fast response and robustness. The lab model on which the case study is going on is dealing with a
2 degrees of freedom tail plane that shows either pitching or yawing motion. In the current study,
pitch control of the tailplane is examined which will done by movement of the elevator that is
attached to a servo motor. To vary the pitching motion of the tail-plane mathematical modeling
and designing of a controller is done along simulation on MATLAB and Simulink. With the help
of the provided data, a transfer function was created which was simulated in an open loop in
Simulink. The result i.e., yaw angle had a steady state error, which is usually the case with open
loop systems because the modeling is not a completely accurate representation of the real
physical system. In order to bring the resulting value to the reference (desired) value, a close
loop feedback system was employed using a PID controller. The use of PID controller in the
close loop response gives the most accurate result by minimizing the response time, overshoot
and steady state error and overcome the shortcoming of open loop response
CONTENTS
List Of Figures .............................................................................................................................................. 4
List Of Tables ............................................................................................................................................... 4
1 Introduction ........................................................................................................................................... 5
2 Literature Review.................................................................................................................................. 6
3 Problem Statement ................................................................................................................................ 7
4 Objectives ............................................................................................................................................. 7
5 Mathematical Model ............................................................................................................................. 7
5.1 Steady state error: ......................................................................................................................... 8
5.2 Transfer Function .......................................................................................................................... 9
5.3 Steady State Error ....................................................................................................................... 11
5.4 SIMULINK Model...................................................................................................................... 12
6 PID Tuning Rules ............................................................................................................................... 12
6.1 Ziegler Nichols method:.............................................................................................................. 14
6.2 Root Locus .................................................................................................................................. 17
7 Results ................................................................................................................................................. 24
8 Discussion ........................................................................................................................................... 25
9 Conclusion .......................................................................................................................................... 25
10 References ....................................................................................................................................... 27
11 Appendix ......................................................................................................................................... 28
LIST OF FIGURES
Figure 1: Aircraft motion in 6 axes ............................................................................................................... 5
Figure 2: Experiment response graph ........................................................................................................... 9
Figure 3: Step Response of second order system ........................................................................................ 11
Figure 4: SIMULINK Model of Closed Loop Response ............................................................................ 12
Figure 5: Result of change in Pi .................................................................................................................. 12
Figure 6: Graphical representation of effect of integral parameter ............................................................. 13
Figure 7: Effect of change in integral parameter ........................................................................................ 13
Figure 8: Result of the derivative parameter ............................................................................................... 14
Figure 9: Ultimate Gain Method Rules ....................................................................................................... 16
Figure 10: System Response at Kp = 20000 ............................................................................................... 16
Figure 11: Root Locus of Iteration #1......................................................................................................... 18
Figure 12: System Response of Iteration #1 ............................................................................................... 19
Figure 13: Root Locus Plot of Iteration #2 ................................................................................................. 20
Figure 14: System Response of Iteration #2 ............................................................................................... 21
Figure 15: Root Locus of Final Iteration .................................................................................................... 22
Figure 16: System Response of Final Iteration ........................................................................................... 23
Figure 17: Response comparison ................................................................................................................ 24
Figure 18: Response Comparison ............................................................................................................... 26

LIST OF TABLES
Table 1: Parameters involved in Mathematical Model ................................................................................. 8
Table 2: Response characteristics comparison............................................................................................ 24
1 INTRODUCTION
This complex engineering problem deals with modelling and pitch control of one degree of
freedom tail plane. Degrees of freedom represents the number of independent planes in which an
object can freely move. For an ordinary aircraft there are a total of 6 degrees of freedom. However,
In this complex engineering problem with a lab-based model of a tail-plane. A tail-plane is a plane
that changes its direction of flight by changing angle of its tail. This is done by movement of either
the rudder or elevator. The lab model we are dealing with is a 2 degrees of freedom tail plane that
shows either pitching or yawing motion.
Aircraft aerodynamic modelling is an imprecise technique based on physics principles, empirical
models and machine identification (SI), and flight and wind tunnel measurement facts. However,
experimental methods are still the preferred method for calculating financial derivatives
management. So obviously, some of these methods are often banned when simulating flying cars.
However, in practice, one or a combination of these technologies will be used and been accepted.
One possible way is to use SI, an experimental technique that has proven to be an excellent tool
for developing complex strategies, in which case it is impossible to obtain perceivable fashion. As
the capacity increases, creating a mathematical version from the first standard becomes an arduous
task. Therefore, more and more researchers are applying SI strategies to accumulate dynamic
fashion on UAVs.
Using SI can also be expected to describe the aerodynamic balance and derivation of the tail rotor
dynamics. However, contrary to mathematical modeling, wind tunnel experiments or actual flight
measurements require correct inputs and outputs. Any of these procedures can be included in the
contract to preserve available equipment models.
For this project, we are dealing with pitch control of this tailplane which will be done by movement
of the elevator that is attached to a servo motor. We will perform mathematical modelling and then
design a controller to vary the pitching motion of the tailplane. The software used for designing
and simulation are MATLAB and Simulink.

Figure 1: Aircraft motion in 6 axes


2 LITERATURE REVIEW
In this project we are looking at control systems approach to study dynamics of an UAV which in
our case is a 1 DOF tail-plane that is limited to only pitch motion for the sake of project’s
simplicity. A tail-plane is an experimental setup placed in front of a wind tunnel used to study the
characteristics of flight dynamics, feedback control approaches and validation of models [1]. The
tail-plane has two different types of mechanisms on its tail used to change its direction which are:
• Rudder: For controlling plane’s yaw motion
• Elevator: For controlling plane’s pitch motion
These two peripherals direct the incoming air from wind tunnel to change the direction of motion.
The model here is evaluated using analytical approach that involves creating a mathematical model
for pitch position in time domain from equation of motion about pitching axis, and then by
experimental approach that includes evaluating model on Simulink. The results of these two
approaches are then compared for validation purposes. The controller should be able to reject
disturbance of any sort and the percentage overshoot should be less than 30% [2]. For creating
mathematical model and software simulation help is taken from studying research papers and by
watching video tutorials.
It is beneficial to increase the model from the first principles. On the other hand, the estimation of
version parameters uses SI in the correct version [3]. This high-precision version is often used for
equipment simulation, verification, quality management, loop and feedback verification, and
system management. In addition, the highly compatible version simplifies the throttle control to
meet the overall performance goals of the device without resorting to esoteric feedback control
paradigms [4].
The device is intended to be used as an experimental validator for studying the principles of flight
dynamics, version verification and unique feedback control strategies. The dynamics of the tail
unit are similar to the dynamics of the longitudinal motion of an airplane without fixed wings.
Therefore, increase the number of models describing the aerodynamic forces and moments acting
on the fuselage [5]. However, the flow phenomenon around a fixed paraglide or airflow profile is
complex; dynamic control and floor deflection control also exacerbate this annoyance. Therefore,
any attempt to mathematically define the device's behaviour alone may not provide a usable
version [6].
3 PROBLEM STATEMENT
Conducting detailed mathematical modeling and pitch control of 1-DOF tail plane.

4 OBJECTIVES
The complex engineering problem is aimed at achieving the following objectives:
1. To develop a detailed mathematical model for pitch position of the tail plane
2. To develop the close loop transfer function between pitch angle and elevator deflection
3. To find the the system gain
4. To design PID controllers in using moderen softwares, having less than 30% of maximum
overshoot and must be able to reject any disturbance.
5. Evaluate the real-time performance on test rig based on the tuned PID gains satisfying desired
performance criteria.

5 MATHEMATICAL MODEL
∑ 𝑀𝑐𝑔 = 𝐼𝑦 𝜃̈

𝑀 = 𝑀𝑜 + ∆𝑀
𝜃 = 𝜃0 + ∆𝜃
If 𝑀𝑜 is equal to zero, then
𝑑2 𝑦
∆𝑀 = 𝐼𝑦 . (𝜃 + ∆𝜃)
𝑑𝑡 2 0
∆𝑀 = 𝐼𝑦 ∆𝜃̈

∆𝑀 = 𝑓𝑛 (∆𝛼, ∆𝛼̇ , ∆𝑞, ∆𝑆𝑒 )


𝛿𝑀 𝛿𝑀 𝛿𝑀 𝛿𝑀
∆𝑀 = . ∆𝛼 + . ∆𝛼̇ + . ∆𝑞 + . ∆𝑞̇
𝛿𝛼 𝛿𝛼̇ 𝛿𝑞 𝛿𝑞̇
If the body coincide with the fized frame at t=0 sec, then

∆𝛼 = ∆𝜃 and ∆𝛼̇ = ∆𝑞 = ∆𝜃̇


After substitution
𝛿𝑀 𝛿𝑀 𝛿𝑀 𝛿𝑀
. ∆𝛼 + . ∆𝛼̇ + . ∆𝑞 + . ∆𝑑𝑒 = 𝐼𝑦 ∆𝜃̈
𝛿𝛼 𝛿𝛼̇ 𝛿𝑞 𝛿𝑑𝑒
The equation can be rewritten as
𝛿𝑀 𝛿𝑀 𝛿𝑀 𝛿𝑀
. ∆𝛼 + . ∆𝛼̇ + . ∆𝑞 + . ∆𝑑𝑒 = ∆𝜃̈
𝛿𝛼. 𝐼𝑦 𝛿𝛼. 𝐼̇ 𝑦 𝛿𝑞. 𝐼𝑦 𝛿𝑑𝑒 . 𝐼𝑦
If
𝛿𝑀 𝛿𝑀 𝛿𝑀 𝛿𝑀
= 𝑀𝛼 , = 𝑀𝛼̇ , = 𝑀𝑞 , = 𝑀𝛿𝑒
𝛿𝛼.𝐼𝑦 𝛿𝛼̇ .𝐼𝑦 𝛿𝑞.𝐼𝑦 𝛿𝑑𝑒 .𝐼𝑦

Therefore, the equation can be rewritten as

𝑀𝛼 ∆𝜃 + 𝑀𝛼̇ ∆𝜃̇ + 𝑀𝑞 ∆𝜃̇ + 𝑀𝛿𝑒 ∆𝛿𝑒 = ∆𝜃̈

Once the equation is rearranged

̇∆𝜃̈ − (𝑀𝛼̇ + 𝑀𝑞 )∆𝜃̇ − 𝑀𝛼 ∆𝜃 = 𝑀𝛿𝑒 ∆𝛿𝑒

Parameters

𝑆𝑦𝑚𝑏𝑜𝑙 𝑃𝑎𝑟𝑎𝑚𝑒𝑡𝑒𝑟
𝛿𝑒 𝑒𝑙𝑒𝑣𝑎𝑡𝑜𝑟 𝑎𝑛𝑔𝑙𝑒 𝑜𝑓 𝑑𝑒𝑓𝑙𝑒𝑐𝑡𝑖𝑜𝑛
𝑞 𝑝𝑖𝑡𝑐ℎ 𝑟𝑎𝑡𝑒
𝛼 𝑎𝑛𝑔𝑙𝑒 𝑜𝑓 𝑎𝑡𝑡𝑎𝑐𝑘
𝛼̇ 𝑟𝑎𝑡𝑒 𝑜𝑓 𝑐ℎ𝑎𝑛𝑔𝑒 𝑜𝑓 𝑎𝑛𝑔𝑙𝑒 𝑜𝑓 𝑎𝑡𝑡𝑎𝑐𝑘
𝜃 𝑝𝑖𝑡𝑐ℎ 𝑎𝑛𝑔𝑙𝑒
Table 1: Parameters involved in Mathematical Model

Expanding it to a Laplace transformation function

∆𝜃̈ − (𝑀𝛼̇ + 𝑀𝑞 )∆𝜃̇ − 𝑀𝛼 ∆𝜃 = 𝑀𝛿𝑒 ∆𝛿𝑒

𝑠 2 × 𝜃(𝑠) − (𝑀𝛼̇ + 𝑀𝑞 ) × 𝑠 × 𝜃(𝑠) - 𝑀𝛼 × 𝜃(𝑠) = 𝑀𝛿𝑒 ∆𝛿𝑒 × 𝑢(𝑠)

𝜃(𝑠)[𝑠 2 − (𝑀𝛼̇ + 𝑀𝑞 ) × 𝑠 - 𝑀𝛼 ] = 𝑀𝛿𝑒 ∆𝛿𝑒 × 𝑢(𝑠)


𝜃(𝑠) 𝑀𝛿𝑒 ∆𝛿𝑒
𝑇(𝑠) = =
𝑢(𝑠) [𝑠2 −(𝑀𝛼̇ +𝑀𝑞 )×𝑠 − 𝑀𝛼 ]

5.1 STEADY STATE ERROR:


𝐸(𝑠) = 𝑢(𝑠) − 𝜃(𝑠)
𝐸(𝑠) 𝜃(𝑠)
=1−
𝑢(𝑠) 𝑢(𝑠)
𝐸(𝑠) 𝑀𝛿𝑒 ∆𝛿𝑒
=1− 2
𝑢(𝑠) [𝑠 − (𝑀𝛼̇ + 𝑀𝑞 ) × 𝑠 − 𝑀𝛼 ]

𝐸(𝑠) [𝑠 2 − (𝑀𝛼̇ + 𝑀𝑞 ) × 𝑠 − 𝑀𝛼 ] − 𝑀𝛿𝑒 ∆𝛿𝑒


=
𝑢(𝑠) [𝑠 2 − (𝑀𝛼̇ + 𝑀𝑞 ) × 𝑠 − 𝑀𝛼 ]

[𝑠 2 − (𝑀𝛼̇ + 𝑀𝑞 ) × 𝑠 − 𝑀𝛼 ] − 𝑀𝛿𝑒 ∆𝛿𝑒


𝐸(𝑠) = × 𝑢(𝑠)
[𝑠 2 − (𝑀𝛼̇ + 𝑀𝑞 ) × 𝑠 − 𝑀𝛼 ]
5.2 TRANSFER FUNCTION
The experiment was conducted in the lab, and the experimental data was provided. Using the data
provided, this will be used to find the values of the coefficient of the mathematical model. To do so, the
values of damping equation and natural frequency of the standard form mathematical model as given
below: -
𝑦̈ + 2𝜉𝜔𝑛 𝑦̇ + 𝜔𝑛2 𝑦 = 𝑏𝑢(𝑡)
Once the values is determined, it will be used to develop the parameters of the transfer function to ensure
similarity with the current figure.

Figure 2: Experiment response graph

First, the value of damping coefficient is determined. This is done by using the following formula
𝜉𝜋

𝑀𝑜𝑠 = 𝑒 √1−𝜉 2

From the graph, the maximum overshoot is 0.133


𝜉𝜋

0.133 = 𝑒 √1−𝜉 2

After solving the equation. The damping coefficient is


𝜉 = 0.5400
To find the natural frequency, the period of oscillations is used. Where
2𝜋
𝑇𝑝𝑒𝑟𝑖𝑜𝑑 =
𝜔𝑛 √1 − 𝜉 2

From the graph, the period of oscillation is 1.80 sec


2𝜋
1.80 =
𝜔𝑛 √1 − 0.54002
The final values of natural frequency will be
𝑟𝑎𝑑
𝜔𝑛 = 4.1050
𝑠
Lastly, the step unit of the function is defined as -10 degrees, hence

𝜔𝑛2 𝑦𝑠𝑠 4.10502 (−11.7)


𝑏= = = 19.72
−10 −10
Hence, the final mathematical model will be
𝑦̈ + 4.433𝑦̇ + 16.85𝑦 = 19.7517𝑢(𝑡)
In our experiment, the transfer function of the model is defined as below

∆𝜃̈ − (𝑀𝛼̇ + 𝑀𝑞 )∆𝜃̇ − 𝑀𝛼 ∆𝜃 = 𝑀𝛿𝑒 ∆𝛿𝑒

𝑠 2 × 𝜃(𝑠) − (𝑀𝛼̇ + 𝑀𝑞 ) × 𝑠 × 𝜃(𝑠) - 𝑀𝛼 × 𝜃(𝑠) = 𝑀𝛿𝑒 ∆𝛿𝑒 × 𝑢(𝑠)

𝜃(𝑠)[𝑠 2 − (𝑀𝛼̇ + 𝑀𝑞 ) × 𝑠 - 𝑀𝛼 ] = 𝑀𝛿𝑒 ∆𝛿𝑒 × 𝑢(𝑠)


𝜃(𝑠) 𝑀𝛿𝑒 ∆𝛿𝑒
𝑇(𝑠) = 𝑢(𝑠)
= [𝑠2 −(𝑀
𝛼̇ +𝑀𝑞 )×𝑠 − 𝑀𝛼 ]

The transfer function of our analysis is defined the following


19.72
𝑇(𝑠) =
𝑠2 + 4.433𝑠 + 16.85
After comparing the coefficients, the values of the co-efficient were compared, where
𝑀𝛼̇ + 𝑀𝑞 = −4.433

𝑀𝛼 = −16.85
Using this, the values of the moment arm were calculated by comparison.
The transfer function was then plotted and compared with the initial figure to ensure similarity.
Figure 3: Step Response of second order system

5.3 STEADY STATE ERROR


The desired steady state value of the system is −10 𝑑𝑒𝑔, meanwhile the system oscillating at the steady
state of approximately −11.7 𝑑𝑒𝑔. To find out the steady state value
10
𝑠. − 𝑠 10(𝑠 2 + 4.433𝑠 + 16.85)
𝑒𝑠𝑠 = lim =−
𝑠→∞ 19.72 𝑠 2 + 4.433𝑠 + 36.57
1+ 2
𝑠 + 4.433𝑠 + 16.85
Using L’Hopital Rule.

10(𝑠 2 + 4.433𝑠 + 16.85) 10(16.85)


lim = − 2
=− = −4.6076
𝑠→∞ 𝑠 + 4.433𝑠 + 36.57 36.57
Next our objective is to reduce the steady state of the value by designing a PID controller.
5.4 SIMULINK MODEL
Once the transfer function was obtained, the SIMULINK model was designed for the control system. The
closed loop feedback was used, where the output was attached back to the input. Furthermore, a PID
controller and the saturation block were also attached to reduce the steady state error.

Figure 4: SIMULINK Model of Closed Loop Response

6 PID TUNING RULES


The proportional term, often called P Constant, can be referred to as Proportional Gain or just Gain,
which is not a unit but instead a ratio. This parameter can also be called Proportional Band and measured
in the unit of percent.
The parameter can be called KP, Gain, or others. This is the parameter that determines how fast the
system responds. The name by which it is referred varies with the manufacturer.
1. For controllers that use the term Gain, adjusting this tuning parameter higher may cause more
sensitive, less stable loops.
2. Conversely, on controllers with proportional band units, decreasing this tuning parameter affects
the loop in the same manner.
Keeping this in mind, knowing the type of controller you have is essential to ensure that you are properly
adjusting your parameters.

Figure 5: Result of change in Pi


PID tuning Integral term
The Integral term or I Constant often called Reset can be expressed differently as well such as:
– repeats per second,
– seconds per repeat,
– repeats per minute,
– minutes per repeat.
Regardless of the measurement type, the integral is the sum of all of the values reported from the signal,
captured from when you started counting to when you completed counting or the area under a plotted
curve. This parameter can be called Ki or Ti. This parameter determines how fast the steady-state error is
removed.

Figure 6: Graphical representation of effect of integral parameter

Because of the different measurements, this parameter may not be as intuitive to adjust. Briefly, smaller
values in minutes per repeat measurements will create larger integral action where larger values in repeats
per minute measurements, will create larger integral action.

Figure 7: Effect of change in integral parameter


PID tuning Derivative term
Derivative or D Constant units are typically seconds or minutes. The purpose of the Derivative constant is
for predicting change. The Derivative action acts of the rate of change measured in the Process Variable.
The value of this parameter basically means how far in the future you want to predict the rate of change.
This parameter can help to create a faster response in your loop and a better performing loop as well.
However, since the Derivative term is measuring the rate of change in the Process Variable, the Process
Variable must be a very clean signal meaning no noise within the signal. For that reason, Derivative terms
are not often used in controls.

Figure 8: Result of the derivative parameter

6.1 ZIEGLER NICHOLS METHOD:


S-curve Method
Implementing a PID controllers requires three independent gains in order to achieve the good close
loop performance. Ziegler Niclos Method provides a good starting point for the selection of these
3 gains. Many dynamic systems exhibits “S-shaped” curve behavior. The reaction curves are
obtained experimentally by applying a step input and measuring the output in an open loop manner.
Figure 9: Ziegler Nichols S-curve Method

Following graph shows the S-curve (red line). From the two rules of Zeigler niclos, the S-curve
Method will be used. The key parameters of S-Curve are delay time Td and slope. Apart from
these, following parameters are extracted:
𝐴 = −13.2
𝑇𝑑 = 0.3 𝑠𝑒𝑐
𝑇 = 0.8 𝑠𝑒𝑐
𝑅 = 16.5 𝑑𝑒𝑔𝑟𝑒𝑒/𝑠𝑒𝑐
So, using the table 10.1, gains for the PID controller are calculated:
1.2 1.2
𝐾𝑝 = = = −0.242
𝑅𝑇𝑑 (−16.5)(0.3)
0.6 0.6
𝐾𝐼 = 2 = = −0.404
𝑅𝑇𝑑 (−16.5)(0.3)2
0.6 0.6
𝐾𝐷 = = = −0.363
𝑅 (−16.5)

Ultimate Gain Method


Tunning the PID controller for required gains is one of the most difficult tasks in controls. Increasing the
proportional gain K_p tends to speed up the response but at the same time may increase the overshoot,
increasing the integral gain K_I its tends to reduce the steady-state error but at the same time may slow
down the speed of response, increasing the derivative gain K_D tends to reduce the overshoot and the
settling time. Implementing a PID controller requires selecting three independent gains to achieve a good
balance in the closed-loop performance criteria such as response time, overshoot, settling time, and
steady-state error. Ziegler Nichols in 1940 developed two methods for selecting “good PID gains”. One
method was the “reaction curve” method where an open loop step response of dynamical system exhibit
“s” shaped curve called the reaction curve. The main parameters of the reaction curve are the delay time
T_D and slope R.
Both parameters are obtained by drawing a line tangent to the inflection point of the S curve. The second
method developed was obtaining marginally stable closed loop response. In this method, the P-controller
gain is continually increased until the closed-loop response transitions from damped sinusoidal
oscillations to sustained oscillations with constant amplitude. therefore, the closed-loop system is
marginally stable. This P-gain setting that results in sustained oscillations called the “ultimate gain” K_U
. The period of the sustained oscillations called the “ultimate period” P_U and is used in the PID tuning.
The ultimate period can be obtained from experiment performed with physical systems. Table 1 represent
Ziegler Nichols tunning rules using the Ultimate gain method.

Controller type Gain


P 𝐾𝑃 = 0.5𝐾𝑈
0.54𝐾𝑈
PI 𝐾𝑃 = 0.5𝐾𝑈 , 𝐾𝐼 = 𝑃𝑈
1.2 𝐾𝑈
𝐾𝑃 = 0.6𝐾𝑈 , 𝐾𝐼 =
PID 𝑃𝑈
𝐾𝐷 = 0.075𝐾𝑈 𝑃𝑈
Figure 10: Ultimate Gain Method Rules

𝐾𝑝 = 20000

Figure 11: System Response at Kp = 20000


Despite increasing the Kp value steadily, no sustained oscillations were observed in our control system.
Therefore, it was concluded that the Ultimate Gain Method was not the appropriate method to deduce the
PID controller parameters.

6.2 ROOT LOCUS


In control theory, root locus method is a graphical technique for finding how the roots of a system change
with the change of a certain system parameter, commonly a gain within a feedback system. This method
used as a stability criterion in the field of control theory and it can determine stability of the system.
The root locus method can be used to design the damping ratio (ζ) and natural frequency (ωn) of a
feedback system instead of just determining the stability of the system. Lines of constant damping ratio
can be drawn radially from the origin and lines of constant natural frequency can be drawn as cosine
inverse whose center points coincide with the origin. By selecting a point along the root locus that
coincides with a desired damping ratio and natural frequency, a gain K can be calculated and
implemented in the controller. More comprehensive methods of controller design using the root locus are
also available and PID controllers can be designed approximately with this technique.
The transfer function for the PID Controller works out to be
𝐾𝐼
𝐺𝑐(𝑠) = 𝐾𝑃 + + 𝐾𝐷 𝑠
𝑠
𝐾𝐷 𝑠 2 + 𝐾𝑃 𝑠 + 𝐾𝐼
𝐺𝑐(𝑠) =
𝑠
Taking 𝐾𝐷 common,
𝐾𝑝 𝐾
𝐾𝐷 (𝑠 2 + 𝐾 𝑠 + 𝐾 𝐼 )
𝐷 𝐷
𝐺𝑐(𝑠) =
𝑠
𝐾𝑃 𝐾𝐼
Considering = 𝐴, =𝐵
𝐾𝐷 𝐾𝐷

𝐾𝐷 (𝑠 2 + 𝐴𝑠 + 𝐵)
𝐺𝑐(𝑠) =
𝑠
𝐾𝐷 (𝑠 + 𝑧1 )(𝑠 + 𝑧2 )
𝐺𝑐(𝑠) =
𝑠
To start off, different iterations will be processed and then the best PID control parameters will be
developed.

Iteration #1
Initially, the roots of the control system is specified as the following: -
𝑧1 = 10 + 5𝑖
𝑧2 = 10 − 5𝑖
Hence the function 𝐺𝑐 (𝑠) is defined by
𝐾𝐷 (𝑠 + 10 + 5𝑖)(𝑠 + 10 − 5𝑖)
𝐺𝑐 (𝑠) =
𝑠
𝐾𝐷 (𝑠 2 + 20𝑠 + 125)
𝐺𝑐 (𝑠) =
𝑠

Figure 12: Root Locus of Iteration #1

Using the locus plot, the value of the gain was determined. This is later substituted as 𝐾𝑑 . Where:-
𝐾𝑑 = 2.46

(2.46)(𝑠 2 + 20𝑠 + 125)


𝐺𝑐 (𝑠) =
𝑠
2.46𝑠 2 + 49.2𝑠 + 307.5
𝐺𝑐 (𝑠) =
𝑠
Equating this with the equation, the values of PID parameters were determined.

𝐾𝐷 𝑠 2 + 𝐾𝑃 𝑠 + 𝐾𝐼
𝐺𝑐(𝑠) =
𝑠
Where
𝐾𝐷 = 2.46
𝐾𝑃 = 49.2
𝐾𝐼 = 307.5
Plotting the response

Figure 13: System Response of Iteration #1

Here, it can be noticed that the overshoot is greater than 30%. Hence these parameters does not fulfill our
requirement.

Iteration #2
Then, the roots of the control system is specified as the following:-
𝑧1 = 3 + 𝑖
𝑧2 = 3 − 𝑖
Hence the function 𝐺𝑐 (𝑠) is defined by
𝐾𝐷 (𝑠 + 3 + 𝑖)(𝑠 + 3 − 𝑖)
𝐺𝑐 (𝑠) =
𝑠
𝐾𝐷 (𝑠 2 + 6𝑠 + 10)
𝐺𝑐 (𝑠) =
𝑠
Figure 14: Root Locus Plot of Iteration #2

𝐾𝐷 = 0.361
(0.361)(𝑠 2 + 6𝑠 + 10)
𝐺𝑐 (𝑠) =
𝑠
0.361𝑠 2 + 2.166𝑠 + 3.61
𝐺𝑐 (𝑠) =
𝑠
quating this with the equation, the values of PID parameters were determined.
𝐾𝐷 𝑠 2 + 𝐾𝑃 𝑠 + 𝐾𝐼
𝐺𝑐(𝑠) =
𝑠
Where
𝐾𝐷 = 0.361
𝐾𝑃 = 2.166
𝐾𝐼 = 3.61
Figure 15: System Response of Iteration #2

The plot fulfill all the responses but the response time is too short which indicates that a lot of energy will
be expended.

Final Iteration
Finally, the roots of the control system is specified as the following: -
𝑧1 = 5 + 𝑖
𝑧2 = 5 − 𝑖
Hence the function 𝐺𝑐 (𝑠) is defined by
𝐾𝐷 (𝑠 + 5 + 𝑖)(𝑠 + 5 − 𝑖)
𝐺𝑐 (𝑠) =
𝑠
𝐾𝐷 (𝑠 2 + 10𝑠 + 26)
𝐺𝑐 (𝑠) =
𝑠
Figure 16: Root Locus of Final Iteration

𝐾𝐷 = 0.163
(0.163)(𝑠 2 + 10𝑠 + 26)
𝐺𝑐 (𝑠) =
𝑠
0.163𝑠 2 + 1.630𝑠 + 4.238
𝐺𝑐 (𝑠) =
𝑠
quating this with the equation, the values of PID parameters were determined.
𝐾𝐷 𝑠 2 + 𝐾𝑃 𝑠 + 𝐾𝐼
𝐺𝑐(𝑠) =
𝑠
Where
𝐾𝐷 = 0.163
𝐾𝑃 = 1.630
𝐾𝐼 = 4.238
Plotting the response of the graph
Figure 17: System Response of Final Iteration
7 RESULTS
The response obtained using the PID controller is plotted in the Figure 18 along with the response we had
without using the PID controller and the plot of system response is there as well.

Figure 18: Response comparison

The system optimized system characteristics were obtained using several techniques including root locus
and Ziegler Nichols reaction curve method. The characteristics of our system obtained with using PID
controller and with open loop response are given in table 2.

Sr. No. Property Name Response With PID Response Without PID
1 Rise time (s) 0.414 1.5189
2 Settling Time (s) 3.336 1.5141
3 Steady State error 0 -4.6076
3 Peak value (Degrees) -11.7 1.32
4 Peak time (s) 0.601 2.0141
5 Overshoot (%) 17 13.32
6 Undershoot (%) 0 0
7 𝐾𝑝 N/A 1.630
8 𝐾𝐼 N/A 4.238
9 𝐾𝐷 N/A 0.163
10 Poles N/A −2.69 + 5.43𝑖
Table 2: Response characteristics comparison
8 DISCUSSION
The results shown in the Figure 18 suggest that use of PID controller in the close loop response gives the
most accurate result by minimizing the response time, overshoot and steady state error and can overcome
the shortcoming of open loop response. Finding the best values of 𝐾𝑝 , 𝐾𝐷 and 𝐾𝐼 ensured the overshoot to
be within the given constraints, which comes out to be 17 %, keeping the system stable. The tracking
error was also reduced by utilizing PID. The one of the most difficult tasks was to obtain these optimized
values of 𝐾𝑝 , 𝐾𝐷 and 𝐾𝐼 due to their non-linear response. Therefore, the trial-and-error method was not
the choice here to find these values. There is no defined methodology for these value selections, however
some guidelines may be followed by optimizing these PID gains to obtain the desired system response.
The auto tuning software called PID Tuner may have been employed, which may offer an automatic
system reaction, but may produce the abnormal gains that cannot be physically attained.

9 CONCLUSION
This report extensively covers the modelling and pitch control of 1-DOF tail plane whose motion is
similar to that of a typical aircraft. The aim of this project is to assess issues in terms of modelling and
controlling of modern unmanned aerial vehicles (UAV) for research purposes. In the For this experiment,
only the pitch motion of the tail plane is considered for control. The tail plane is placed in front of the
wind tunnel and it changes its pitch motion by using the elevator part of its tail by directing the incoming
air from wind tunnel in the desired direction. In this project we have successfully demonstrated that how
can vary the pitch motion of tail plane by designing and integrating a controller. First task was to create
mathematical model for pitch position in time domain using aircrafts equation of motion for pitch motion.
Pitch plane model parameters were provided by the instructor that were used alongside the equation of
motion (EOM) for creating mathematical model. The standard form of mathematical model used for pitch
control is:

𝑦̈ + 2𝜉𝜔𝑛 𝑦̇ + 𝜔𝑛2 𝑦 = 𝑏𝑢(𝑡)


After this, transfer function was formed from calculating the values such as damping coefficient, natural
frequencies etc. The steady state error was calculated to be -4.6076 which is due to open loop system
having a difference with real world systems. This was minimized by implementing a closed loop system
approach that involved designing PID controller with several iterations. The closed loop transfer function
is between the elevator deflection and pitch angle. Finally, after various attempts the appropriate design of
PID controller was selected with a set of gain values that work to minimize error and they are stated as:

• Proportional gain (Kp): 1.630


• Integral gain (Ki): 4.238
• Derivative gain (Kd): 0.163

The plot obtained from these was also suitable for design (Overshoot < 30%, Minimum energy losses).
The optimized system parameters were found using Ziegler Nichols reaction curve method and the root
locus method. These parameters are discussed in the result portion of the report in form of a table and
they satisfy the desired characteristics of the system (e.g. Overshoot = 17 %).
One of the main objectives of this project was to model the pitch dynamic system as much close to real
world results as possible. This essentially means to create a high-fidelity system and then validate it by
means of experimentation. A comparison between analytical results (Mathematical model) and
experimental results (software results) was performed in terms of a comparison plot.

Figure 19: Response Comparison

The above response comparison plot shows three response curves shown in different colors after applying
a step input of -10 :

• The blue line represents system output that was obtained after performing experiments
• The red line shows response without implementing PID controller. This was done by equating
system output with transfer function
• The yellow line shows the response with PID controller implemented

Here we can observe that the response of experimental data (blue line) is very close to the mathematical
model (red line). This shows that we have successfully modelled a high-fidelity system. The slight
deviation between the experimental and mathematical response is due to energy dissipation factors like
bearing fiction in the apparatus which was ignored while modelling the system to avoid any complexity in
the design process. In future, for improvement of model better results we can design a mathematical
model that incorporates friction effects which actually occur in real world environment. Thus, due to low
distortion and having the desired parameters, this project is considered successful and ready to be
launched in real system implementations.
10 REFERENCES
[1] Ahmad, S. M. (2013). Flight dynamics, parametric modelling and real-time control of a 1-DOF
Tailplane. Mathematical and Computer Modelling of Dynamical Systems, 19(3), 220–
237. doi:10.1080/13873954.2012.724427
[2] Robert C. Nelson, Flight Stability and Automatic Control, 1998
[3] Khan, Y. M., Ahmad, S. M., Ali, M., & Khan, M. (2019). Flight dynamics and parametric
modeling of a 2-DOF lab aircraft. Proceedings of the Institution of Mechanical Engineers, Part
G: Journal of Aerospace Engineering, 233(8), 2923-2931.
[4] Ezekiel, D. M., Samikannu, R., & Matsebe, O. (2021). Pitch and yaw angular motions
(rotations) control of the 1-DOF and 2-DOF TRMS: A survey. Archives of Computational
Methods in Engineering, 28(3), 1449-1458.
[5] Chaudhary, S., & Kumar, A. (2019, June). Control of Twin Rotor MIMO system using 1-
degree-of-freedom PID, 2-degree-of-freedom PID and fractional order PID controller. In 2019 3rd
International conference on Electronics, Communication and Aerospace Technology
(ICECA) (pp. 746-751). IEEE.
[6] Pattinson, J., Lowenberg, M. H., & Goman, M. G. (2013). Multi-degree-of-freedom wind-
tunnel maneuver rig for dynamic simulation and aerodynamic model identification. Journal of
Aircraft, 50(2), 551-566.
11 APPENDIX
Transfer Function and System Response Plot
close all;
clear all; clc;
open('G5_Output_plot_input_Elevator_-10degree.fig')
a=get(gca,'Children');
xdata=get(a,'Xdata');
ydata=get(a,'Ydata');
% plot(xdata,ydata);
wn=4.105;
z=0.540;
b=19.7157;
num=[b];
den=[1 2*z*wn wn^2];
sys=tf(num,den)
t=0.3:0.05:25.3;
u=-10*ones(size(t));
figure(1)
hold on;
[y,t] = lsim(sys,u,t)
plot(t,y)
title("Step response of the second order, underdamped undamped system");
xlabel("Time (s)");
ylabel("Elevator angle of Deflection (degrees)");
legend('System Response','Transfer function plot')

Root Locus Plot


close all;
clear all;
clc;

sysGH = tf([19.72],[1 4.433 16.85]);


GH = tf([1 10 26], [1 0]);
rlocus(sysGH*GH);
sgrid

System Response with and without PID control graphs


clear all;
close all;
close all;
time=0.3:0.05:25.3;
open('G5_Output_plot_input_Elevator_-10degree.fig')
a=get(gca,'Children');
xdata=get(a,'Xdata');
ydata=get(a,'Ydata');
plot(xdata,ydata);
hold on
format long;
sim closedloop
x1 = non_pid.Data()'
plot(time,x1(1:501))
hold on
x2 = pid.Data()'
plot(time,x2(1:501))
xlabel('Time (s)')
ylabel('Yaw Angle (degree)')
title('Response Comparision')
xlim([0,25])
grid
legend('System Output','Without PID Controller','With PID Controller')

Properties of the open loop response


wn=4.105;
z=0.540;
b=19.7157;
num=[b];
den=[1 2*z*wn wn^2];
sys=tf(num,den)
t=0:0.05:25;
u=-10*ones(size(t));
figure(1)
hold on;
[y,t] = lsim(sys,u,t)
xlim([0 5])
grid
plot(t,y)
s= stepinfo(sys)

View publication stats

You might also like