Sie sind auf Seite 1von 10

School of Engineering

ASSIGNMENT COVER SHEET

Student Name:....................................................................
Student ID:..........................................................................
Major: .......................................................................... ......
Year of study: .....................................................................
Lecturer:.............................................................................
Personal Tutor:..................................................................
Subject:...............................................................................
Assignment Title/Number................................................

I declare that all material in this assignment is my own work except where there is clear
acknowledgment or reference to the work of others and I have read the University statement on
Academic Misconduct (Plagiarism) in the Student Handbook.

Signed:.............................................................. Date:..............................................................

Problem 1.
To solve this problem MATLAB software has been used. My ID number is
201493886. Last four digits 3886.
K = 3886*9/10000=3.4974
Transfer function is equal to:

2
s( s+ 1)(s+ 4)
2K
6.9948
T ( s )=
=
= 3
2
2
s ( s+1 ) ( s +4 ) +2 K s +5 s + 4 s +6.9948
1+ K
s ( s+1)(s +4)
K

This code has been used in MATLAB:


b = [0 0 0 6.9948];
a = [1 5 4 6.9948];
fvtool(b,a,'polezero')
[b,a] = eqtflength(b,a);
[z,p,k] = tf2zp(b,a)
text(real(z)+.1,imag(z),'Zero')
text(real(p)+.1,imag(p),'Pole')
Output is:

Figure 1. Location of poles on s-plane


z = Empty matrix: 0-by-1
p = -4.4545 + 0.0000i
-0.2727 + 1.2231i
-0.2727 - 1.2231i
From the graph it can be observed that all poles are located on LHP, this means
that the system is stable.
Problem 2.
In this task we have two input in our system and each of them can contribute to
steady state error. In calculation this two inputs can be considered separately.

Y ( s)
2K
=
R ( s ) S ( s+ 1 )( s+ 4 ) +2 K
So when s goes to 0 steady state error for step input is zero independently from
K.
Second inputs (disturbance) error is calculated in next way:

Y (s)
2
=
T D ( s) s ( s +1 ) ( s+ 4 )+ 2 K
For step disturbance input Td(s) = A/s:

lim s 0 sE ( s )= A / K

The condition for this task is to minimize the error so K should be big as much as
possible. It was given that K maximum can be 10 in order to keep the system
stable. So in our design for the value of K 10 has been chosen.

Problem 3.
In this question PID parameters should be chosen in order to obtain a system
with overshoot less 5%, also Ts 2 secconds. In this task also MATLAB has been
used.

G ( s )=

10
s (s +2)(s+4)

To command window just the transfer function has been written and simple code
pidtool automatically makes tuning process.
Input:
f = tf([0 0 0 10],[1 6 8 0]);
pidtool(f,p)

Output:

Figure 2. The system with P controller


The Kp variable can be adjusted using the roller but it has been observed that for
any number Kp the initial conditions can not be satisfied. So next PI controller
has been checked:
Input:
pidtool(f,pi)
Output:

Figure 3. The system with PI controller.


In this case it seen that jointly effect of Kp and Ki makes the overshoot very high,
so it doesnt satisfy conditions.
Next step is usage of PID contoller.
Input:

pidtool(f,pid)
Output:

Figure 4. The system with PID contoller


From the diagram above it can be examined that settling time is lower than 2
and overshoot is also satisfy initial criterions.

So final values are:


Kp = 1.9542; Ki = 0.089114; Kd = 1.4641.
Peak time, Tp, is 1.03 s. Also Tr = 0.526. The blue dot at the graph above
represents steady state error. From the the final value is 1, which means that e ss
is zero.
Problem 4
In this question to calculate parameters for contoller Ziegler-Nichols tuning
method has been. We have to approach to this problem:
1. The value for Ku and Pu are calculated in following way:
s3
s2
S

1
6

s0

10Kp

6 810 Kp
6

8
10Kp
0
0

10Kp > 0 or Kp > 0, and

6 810 Kp
>0
6
68 10Kp>0
4.8 > Kp
So it means that Ku is equal to 4.8.
(jw)3 + 6(jw)2 + 8(jw) + 10Kp = 0
-w3j 6w2 + 8wj + 10Kp = 0
1) 10Kp = 6w2
w = 2

2) w3 = 8w
w = 0, 2

Pu = 2/w = 2.22 s

So Pu is equal to 2.22 s. Next standart table can be constructed:

Type of Controller
P
PI

General Formula
Kp = 0.5Ku
Kp = 0.45Ku

Optimum Gain
Kp = 2.4
Kp = 2.16

Parameters
Kp = 2.4
Kp = 2.16

PID

TI = Pu/1.2
Kp = 0.6Ku

TI = 1.85
Kp = 2.88

KI = 1.17
Kp = 2.88

TI = 0.5Pu

TI = 1.11

KI = 2.59

TD = 0.125Pu

TD = 0.278

KD = 0.80

2) the second method is configuring the Kp until the system becomes oscillatory
and this value is considered as Ku. Code:
Kp=4.8;
Ki=0;
Kd=0;
C=pid(Kp,Ki,Kd);
T=feedback(C*f,1);
step(T)

Output:

Figure 5. The system with adjusted Kp for ocsillation

The period can be obtained from the difference between two points at the top.
From this method it can be confirmed that Ku is 4.8 and Pu is 2.22. Next steps are
identical to previous one.
Conclusion
In this project we analyzed PID contoller and some tuning methods. Application
MATLAB greatly easened the work and proved to be very useful tool for designing
PID. From third tasks results show that implementing only P or PI controllers can
not satisfy initial conditions, which can be explained with their jointly effect on
the overshoot, both of them increase it. By using software, PID can be easely
adjusted to any parameter, changing the variables for settling time and rise
time.
In fourth task Ziegler-Nichols tuning method was used in two different ways. If we
compare this values with the data on 3rd problem, some divergences can be
observed. But it should be kept in mind the Ziegler-Nichols method gives only
estimate numbers.

Das könnte Ihnen auch gefallen