Prepared by

Abeer ALNuaimi
Balqees ALDaghar
Afra Ebrahim
Lila Abdullah
United Arab Emirates University
College of Engineering
Electrical Engineering Department
Project Advisor
Dr. Abdulla Ismail
200203257
200324560
200310882
Contents
Introduction
Summary about our project.
Review GP1 task.
Gantt chart for GP2
optimal control:
LFC with PI & optimal control
AVR with PI & optimal control
Contents
Combination of LFC and AVR
LFC with Fuzzy logic control
LFC with Robust control
Comparison for three controllers
PI, Fuzzy & Robust
Conclusion
AGC Overview
The system:
Power Generation system.
The problems:
Frequency and voltage variations
The consequences:
Machine damage.
Blackouts, or outages.
GP1 Overview
The Project:
Automatic Generation Control system
The Advantages:
Limits the variations.
Avoide machine damages
Avoide blackouts
Enhance the system reliability and security.


GP1 Overview
GP1
Model
LFC
system
One area
Integral
control (IC)
SIMULINK
model- block
diagrams
MATLAB
code
Two
areas
Three
areas
GP1 Overview



Gp1
Gantt chart GP2 Plan


Gantt chart GP2 Plan
Optimal Linear Control Systems
optimal control is a set of differential equations describing the
paths of the control variables that concerned with operating a
dynamic system to minimize the cost functional with weighting
factors supplied by a engineer.
Example for optimal control
Optimal Linear Control Systems
Application of optimal control:
Mechanics of motion.
Economics.
Medicals.
Populations.

The targets for using the optimal linear
control system:
1. Stable closed-loop system.
2. Reduce steady state errors.
3. Reach standard performance
measures:
Peak Time, Tp.
Percent of overshoot.
Percent of under shoot.
Settling time, Ts.
Rise time, Tr.

Minimization cost equation:
| | 0 , 0 ) ( ) ( ) ( ) (
2
1
2 1 2 1
) (
min
> >

+ =
}
R R dt t u R t u t x R t x J
if
io
T T
t u
State variable
Input
The LFC with the I and OPC
l
f1
0.875
s
control l er
1
0.3s+1
Turbi ne
PL
1
0.08s+1
Governor
1.428
0.37s+1
Generator
0.4
Gai n
Model1: With the integral control.
The LFC with the I and OPC
MATLAB: Defining the Matrices.
A=[-12.5 0 -12.5 -5;3.33 -3.33 0 0;0 3.86 -2.70 0;0 0 0.87 0];
B=[12.5;0;0;0]; F=[0;0;-1.93;0];
C=[0 0 1 0];
D=[0];

The LFC with the I and OPC

0 5 10 15 20 25 30
-0.045
-0.04
-0.035
-0.03
-0.025
-0.02
-0.015
-0.01
-0.005
0
Model1 output respronse without integral controller
Time (s)
d
f
/
d
t

(
H
z
)
US: 4%
SSR: t=25s
The LFC with the I and OPC
Model1: With the integral control and optimal control.
f2
f1
-0.4s-0.875
s
control l er1
-0.875
s
control l er
1
0.3s+1
Turbi ne1
1
0.3s+1
Turbi ne
PL1
PL
1
0.08s+1
Governor1
1
0.08s+1
Governor
1.428
0.37s+1
Generator1
1.428
0.37s+1
Generator
-K-
Gai n4
-K-
Gai n3
-K-
Gai n2
-K-
Gai n1
-K-
Gai n
du/dt
Deri vati ve
-K-
-1.143
The LFC with the I and OPC
MATLAB: Defining the Matrices.
Q=[10 0 0 0; 0 10 0 0 ;0 0 10 0 ;0 0 0 10];%
R=1;
[K,P,ev]=lqr(A,F,Q,R)
Ao=A-(F*K)
sys1=ss(Ao,F,C,D);
yo=lsim(Ao,F,C,0,u,t);
The LFC with the I and OPC

0 5 10 15 20 25 30
-0.045
-0.04
-0.035
-0.03
-0.025
-0.02
-0.015
-0.01
-0.005
0
Model1 output respronse without integral controller(Blue) and Optimal control(green)
Time (s)
d
f
/
d
t

(
H
z
)
US: 1.5%
SSR: t=10s
The LFC with the I and OPC
The Integral and the Optimal control Advantages:
Undershoot Reduction from 4% to 1.5%
The steady state response deducted faster
The integral control helps in enhancing the steady state response
from t=25s to t=10s.
The Optimal control helps in enhancing the Transient response.

System Models
(AGC)
Valve Control
mechanism
Load frequency
control (LFC)
Frequency
sensor
Voltage sensor
Automatic Voltage
Regulator (AVR)
Excitation
system
Gen. field
Turbine
G
P A
Steam
tie
P A
V
P A
G G
Q P A A ,
G
Shaft




Automatic Voltage Regulation
For efficient and reliable operation of Power Systems,
the control of voltage should satisfy the following
objective:

Voltages at the terminals of all equipment in the system are
within acceptable limits. Maintaining voltages within the
required limits is complicated due to the fact that:

1) The power system supplies power to vast number of loads and fed from
many generating units.



2) System voltage is closely related to the system reactive power
which is a reactive loads such as inductors and capacitors dissipate
zero power, yet the fact that they drop voltage and draw current
gives the deceptive impression that they actually do dissipate
power.

3) The proper selection and coordination of equipment for controlling
the system voltage and the reactive power are among the major
challenges in power system operation and control.

Automatic Voltage Regulation
Reactive Power (QV) is one of the two main elements in
the power system must be controlled.

Any voltage error in the system is sensed, measured,
and transformed into reactive-power command signal.

The objective of the AVR is to keep the system terminal
voltage at the desired value by means of feedback
control

Automatic Voltage Regulation
Block diagram of AGC model
AVR Model
Block diagram of a simple automatic
voltage regulator (AVR)

K
E
=200 T
E
=0.05
K
G
=1 T
G
=0.2
K
R
=0.05 T
R
=0.05
KA=0.15 TA=10
Voltage error is improved by controlling the rotor field-current
generator EMF.
The steady state voltage error can be eliminated using an
integral controller.
The AVR has a substantial effect on transient stability when
varying the field voltage to maintain the terminal voltage
constant.
Block diagram of a simple automatic
voltage regulator (AVR)

31
AVR Model

Case 1:
AVR without PI (Proportional and Integral ) controller.

Case 2:
AVR with PI controller.

Case 3:
AVR with optimal control.
Case 1: AVR without PI (Proportional and
Integral ) controller.
Block diagram of AVR model without PI controller
The output voltage response without
controller
Steady State error
Overshoot error
Time (s)
V
Overshoot
Steady state error
The output voltage response when Ka
of the amplifier was changed to 0.1
The output voltage response when Ka
of the amplifier is 0.15
Case 2: AVR with PI controller.


Block diagram of AVR model with K
i
and K
p
gains
The output voltage response when Ki=0.2 and Kp= 1.5
Overshoot
Time (s)
V
The output voltage response with PI
controller
Case 3: AVR with optimal control.

Block diagram of AVR model with feedback gains












4 ) ( x s U
3
x
) ( 01 . 0 01 . 0 1 . 0 3
4 3
t u x x x + =
2
x
3 2
4000 20 2 x x x + =
3
x
2
x
1
x
2 1 1
5 5 x x x + =
4
x
1 4 4
20 20 x x x + =
1
x
Step1: Find the state variables and output equations:

State differential Equation:



Output Equation:



A=[-5 5 0 0; 0 -20 4000 0; 0 0 -0.1 -0.01;20 0 0 -20]
B=[0;0;0.01;0]
C=[1 0 0 0]
D=[0]

) ( ) ( ) ( t u B t x A t x
m n n n
+ =
) ( ) ( ) ( t u D t x C t y
m p n p
+ =
Step2: Find A,B, C, D matrices

Step 3: MATLAB command to find the feedback gains

MATLAB command

A=[-5 5 0 0; 0 -20 4000 0; 0 0 -0.1 -0.01;20 0 0 -20]
Q=[5 0 0 0; 0 5 0 0; 0 0 5 0; 0 0 0 5]
B=[0;0;0.01;0]
R=5
[F,P,ev]=lqr(A,B,Q,R)

Result of running the program:

F =

-0.0230 0.0582 206.0278 -0.0899

P =

1.0e+005 *

0.0000 0.0000 -0.0001 0.0000
0.0000 0.0000 0.0003 0.0000
-0.0001 0.0003 1.0301 -0.0004
0.0000 0.0000 -0.0004 0.0000

ev =

-20.6225 + 3.4261i
-20.6225 - 3.4261i
-2.9577 + 3.1290i
-2.9577 - 3.1290i


values of feedback gains
k
1
,k
2
,k
3
,k
4


MATLAB Function
The output voltage response with optimal and integral control

Time (s)
V
AVR
LFC
AGC system
x7 x6 x5
x9
x8
x1 x2 x3
x4
df/dt
Vref (s)
V(t) + df/dt
V(t)
1
0.3s+1
Turbine
1
0.05s+1
Sensor
PL
1.5
Kp
0.2
s
KI
1
K5
0
K4
0.1
K3
1
K2
0.8
K1
0.875
s
Integral control
1
0.08s+1
Governor
1
0.2s+1
Generator1
1.428
0.37s+1
Generator
200
0.05s+1
Exciter
0.1
10s+1
Amplifier
0.4
1/R
x3
Load Frequency Control
Auto Voltage Regulator
AVR and LFC Combination
Forming A, B, C, D and F Matrices:
State Differential Equation.
Output Equation.
MATLAB.
Tuning K1,K2,K3,K4 and K5 Between 0 and 1:
Trial and Error:
K1 has no affect on either one of the two systems.
K2 has an affect on the LFC response.
K3 has an affect on both the LFC and the AVR system
stability.
K4 and K5 both have an affect on the AVR overshoot.

AVR and LFC Combination
Tuning K1,K2,K3,K4 and K5 Between 0 and 1:
K at which the responses of both AVR and LFC are behaving normally:
K1= 1
K2= 0.8
K3= 0.1
K4= 0
K5= 1
AVR and LFC Combination
0 5 10 15 20 25 30
-0.03
-0.025
-0.02
-0.015
-0.01
-0.005
0
0.005
Model1(reordered) output respronse with the integral controller
Time (s)
d
f
/
d
t

(
H
z
)
LFC response
AVR response
AGC response
0 100 200 300 400 500 600 700 800 900 1000
0
20
40
60
80
100
120
x7 x6 x5 x9
x8
x1 x2 x3
x4
df/dt
Vref (s)
V(t) + df/dt
V(t)
1
0.3s+1
Turbine
1
0.05s+1
Sensor
PL
0
Kp
0.09
s
KI
-K-
K9
-K-
K8
-K-
K7
-K-
K6
1
K5
0
K4
0.1
K3
1
K2
-K-
K10
0.8
K1
0.875
s
Integral control
1
0.08s+1
Governor
1
0.2s+1
Generator1
1.428
0.37s+1
Generator
200
0.05s+1
Exciter
0.1
10s+1
Amplifier
0.4
1/R
x3
Load Frequency Control
Auto Voltage Regulator
AVR and LFC Combination
0 5 10 15 20 25 30
-0.03
-0.025
-0.02
-0.015
-0.01
-0.005
0
0.005
Model1(reordered) output respronse with the integral controller
Time (s)
d
f
/
d
t

(
H
z
)
LFC response
AVR response
AGC response
0 50 100 150 200 250 300 350 400 450 500
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Model1 output respronse with integral and optimal controllers
Time (s)
V
o
l
t
s

(
v
)
0 1 2 3 4 5 6 7 8 9 10
-0.03
-0.025
-0.02
-0.015
-0.01
-0.005
0
0.005
Model1y
v
and y
d
f output respronse in an AVR & LFC combination
Time (s)
V
o
l
t
/
D
f
/
d
t

(
v
s
)
0 5 10 15 20 25 30
-0.03
-0.025
-0.02
-0.015
-0.01
-0.005
0
0.005
Model1(reordered) output respronse with the integral controller
Time (s)
d
f
/
d
t

(
H
z
)
LFC response
AVR response
AGC response
0 50 100 150 200 250 300 350 400 450 500
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Model1 output respronse with integral and optimal controllers
Time (s)
V
o
l
t
s

(
v
)
The Combination of both AVR and LFC systems might cause
slight changes in their responses.

Fortunately the undershoots and overshoots never exceeded
20%.
AVR stand alone system
Overshoot = 2.2%
With the optimal control the overshoot almost eliminated.
AVR within AGC system
Overshoot= 0.2%
LFC stand alone system
Undershoot= 2.8%
LFC within AGC system
Undershoot= 2.5%
AVR and LFC Combination
Fuzzy logic control
It is the process of formulating the mapping from a given input to an
output using fuzzy logic. The mapping of FL done based on human
operators behavior.

Fuzzy logic control
Fuzz
y
logic
Natural
language
s
Model
nonlinear
function
Cheaper
Faster
Flexible
Easy to
understand
Fuzzy Logic Control Application
Automatic control
Data classification
Decision analysis
Expert systems
Computer vision
Cameras
washing machines
microwave ovens
Industrial process control
Medical instrumentation

Fuzzy logic control process




Fuzzify Inputs
Apply Fuzzy Operator
Apply Implication Method
Aggregate All Outputs
Defuzzify
Modeling one area LFC with Fuzzy
logic control

FIS Editor

Membership Function

FIS variables
de\e LN MN SN VS SP MP LP
LP VS SP MP LP LP LP LP
MP SN VS SP MP MP LP LP
SP MN SN VS SP SP MP LP
VS MN MN SN VS SP MP MP
SN LN MN SN SN VS SP MP
MN LN LN MN MN SN VS SP
LN LN LN LN LN MN SN VS


FIS variables
Variables Linguistic Term Range of linguistic term
Near +0.05 LP (large positive) [+0.04 +0.05 +0.05]
Near -0.05 LN (large negative) [-0.05 -0.05 -0.04]
So far from +0.05 MP (medium positive) [+0.025 +0.035 +0.045]
So far from -0.05 MN (medium negative) [-0.045 -0.035 -0.025]
Very far from +0.05 SP (small positive) [+0.005 +0.015 +0.03]
Very far from -0.05 SN (small negative) [-0.03 -0.015 -0.005]
f=0 VS (very small) [-0.007 0 +0.007]
Rule Editor

If-And-Then rules
The response of LFC one area
Fuzzy control

Modeling two area LFC with
Fuzzy logic control

Area Control Error
Area control error is the difference between the actual
power flow out of area, and scheduled power flow. ACE
also includes a frequency component.
f f f
P P P
P P P
sch
tie
sch
tie tie
g
sch
g g
= A
= A
= A
The response of LFC two areas
Fuzzy control

F1
F2
The response of Tie line for LFC two
areas Fuzzy control

LFC Model with Robust controller

What is Robust control ?
Why we need Robust control in our model (AGC)?
Applications of Robust control
Robust controller design

Robust controller
The dynamic behavior of electric power systems is heavily
affected by disturbances and changes in the operating points.

An industrial plant such as power systems always contains
parametric uncertainties.

In many control applications, it is expected that the behavior of
the designed system will be insensitive (robust) to external
disturbance and parameter variations

Applications of Robust control


Robust control of Temperature
Disk drive read system
Mobile ,Remote-Controlled video camera
Spacecraft
Control of a(Digital audio tape) DAT player
Elevator
Microscope control


Robust controller design

P
d
(t) : load disturbance (P.u. MW)
T
g
: governor time constant (s)
K
g
: governor gain
T
t
: turbine time constant (s)
K
t
: Turbine gain
T
p
: Generator time constant (s)
K
p
: Generator gain
R : speed regulation due to governor action (HZ
p.u. MW
-1
)
KI: Integral control gain

LFC Block diagram of power system
Our robust load-frequency controller design procedure is
as follows:

Step 1: Find the range of the system parameters

State equation:


Output equation:


Where :
) ( ) ( ) ( t u D t x C t y
m p n p
+ =
)] ( ), ( ), ( ), ( [ ) (
4 3 2 1
t x t x t x t x t x =
) ( ) ( ) ( ) ( t P F t Bu t Ax t x
d
A + + =

1 4
x K x
I
=
1
x
1 4 3
) ( 3
1
x
R T
k
t u
T
k
x
Tg
kg
x
T
x
g
g
g
g
g
+ =
3 2 2
1
x
T
k
x
T
x
t
t
t
+ =
d
P
P
P
P
P
P
T
k
x
T
k
x
T
x A + =
2 1 1
1

A =
-1/Tp Kp/Tp 0 0
0 -1/Tt 1/Tt 0
-1/RTg 0 -1/Tg -1/Tg
K 0 0 0

The range of the system parameters is:


| | 0 , / 1 , 0 , 0
G
T B =
| | 0 , 0 , 0 , /
P P
T K F =
] , [ / 1
] , [ / 1
] , [ / 1
] , [ /
] , [ / 1
5 5
4 4
3 3
2 2
1 1
a a RT
a a T
a a T
a a T K
a a T
G
G
T
P P
P
e
e
e
e
e
] 639 . 10 , 081 . 3 [ / 1
] 857 . 17 , 615 . 9 [ / 1
] 762 . 4 , 564 . 2 [ / 1
] 12 , 4 [ /
] 1 . 0 , 033 . 0 [ / 1
e
e
e
e
e
G
G
T
P P
P
RT
T
T
T K
T
Step 2: Choose the nominal parameters for the system and
decide the bound of the uncertainties

The nominal parameters are from the original model of LFC:







And ,











) ( ) ( ) ( ) ( ) ( t u B B t x A A t x A + + A + =
A A A A + = B B B A + =
A =
-2.7030 3.8595 0 0
0 -3.3330 3.3330 0
-31.2500 0 -12.5000 -12.5000
0.8800 0 0 0

B=[0;0;12.5;0]

F=[3.8595;0;0;0]




Now, let decide the bound of the uncertainties:





Hence, the parametric uncertainties are:


A A o = A
B B | = A
F F = A
7 . 0 & 5 . 0 , 3 . 0 = = = | o
A =
-0.8109 1.1579 0 0
0 -0.9990 0.9999 0
-9.3750 0 -3.7500 -3.7500
0.2640 0 0 0

B=[0;0;6.25;0]

F=[2.70165;0;0;0]


After this change in the system the new matrices are as follow:


A A A A + = B B B A + =
A =

-3.5139 5.0174 0 0
0 -4.3320 4.3320 0
-40.6250 0 -16.2500 -16.2500
1.1440 0 0 0

B=[0;0;18.75;0]

F=[6.56115;0;0;0]
Step 3: Choose the design constants and the design constant
matrices Q and R








And because the algebraic Riccati equation is nonlinear equation we use
MATLAB program to solve it.
( ) ) 1 ( .... 0
1
1
2
1
1
1
Eq Q U P T B R B P P A A P
T T
= + +
)
`

+

c c
c
o
c
Where , Q > 0 and R > 0
&
1
> 0 , very small value
Algebraic Riccati equation:
T & U are the rate change of the generation
Step 4: Use the algorithm given eq. (1) to solve Riccati equation
and obtain the solution P
By using the command from the MATLAB we can found P as
follow:
MATLAB Command

A=[-3.5139 5.0174 0 0;0 -4.332 4.332 0;
-40.625 0 -16.25 -16.25; 1.144 0 0 0]
Q=[5 0 0 0; 0 5 0 0; 0 0 5 0; 0 0 0 5]
B=[0;0;18.75;0]
R=0.01
[F,P,ev]=lqr(A,B,Q,R)

Result of the command

F =
9.6166 17.6707 21.6925 21.5108
P =
1.1042 0.6104 0.0051 1.7927
0.6104 0.9237 0.0094 1.1636
0.0051 0.0094 0.0116 0.0115
1.7927 1.1636 0.0115 7.7951
ev =
1.0e+002 *
-4.1956
-0.0522 + 0.0168i
-0.0522 - 0.0168i
-0.0083

By using MATLAB the output frequency response was drawn
without considering the feedback gains:
MATLAB Command
clc
t=[0:0.1:20];
u=-0.1*ones(length(t),1);
x0=[0 0 0 0];
A=[-3.5139 5.0174 0 0;0 -4.332 4.332 0; -40.625 0 -
16.25 -16.25; 1.144 0 0 0];
eig(A)
B=[0;0;18.75;0];
C=[1 0 0 0];
D=[0];
sys=ss(A,B,C,D);
[y,x]=lsim(sys,u,t,x0);
plot(t,y)
Title('The output frequency respronse');
xlabel('Time');
ylabel('f');
grid


0 2 4 6 8 10 12 14 16 18 20
-0.05
-0.04
-0.03
-0.02
-0.01
0
0.01
The output frequency respronse
Time(s)
D
e
l
t
a

f

(
H
Z
)
The output frequency response with uncertainties parameters and without
feedback gains
Step 5: Construct the feedback gain

Also, by using the same command from MATLAB we found the
optimal gains:


F =
9.6166 17.6707 21.6925 21.5108


LFC Block diagram of power system for the proposed
robust controller

The output frequency response for the proposed robust controller
Comparison between robust and integral controller

With proposed robust controller
With Integral controller
With proposed robust
controller
With Integral controller
Figure 1: With nominal parameters 1/T
p
= 2.7030, K
p
/T
p
=
3.8595, 1/T
T
= 3.333, 1/T
G
=12.5, 1/RT
G
= 31.25, K
I
= 0.88
Figure 2: With 1/T
p
=1.05, K
p
/T
p
=1.494, 1/T
T
=1.3,
1/T
G
=1.79,1/RT
G
=0.7143,K
I
=1.144
With proposed robust
controller
With Integral controller
Figure 3 : With 1/T
p
= 0.8, K
p
/T
p
= 1.1424, 1/T
T
=1.031,
1/T
G
=1.52,1/RT
G
=0.61,K
I
=0.88
With proposed robust
controller
With Integral controller
Figure 4: With 1/T
p
= 0.033, K
p
/T
p
= 4, 1/T
T
=2.564,
1/T
G
=9.615,1/RT
G
=3.081,K
I
=0.88
Comparison
The use of the PID algorithm for control does not guarantee
optimal control of the system or system stability, thats why
in our designs of LFC and AVR we used the optimal linear
control systems.

The Fuzzy-logic controller can be seen as a heuristic and
modular way of defining nonlinear system but the fuzzy
logic controller failed in considering the uncertainties.




Comparison
The proposed robust controller is simple, effective and can
ensure that the overall system is asymptotically stable for
all admissible uncertainties.



Conclusion
Our goal in the end is to design a control system that serves
the power network in the UAE for better performance and
better power services in terms of consumption and
supplement.

Enhance our skills and understanding of Engineering project
design and management.

Achieve the best as an outcome of a successful group work.