Академический Документы
Профессиональный Документы
Культура Документы
II.
AVR MODEL
I. INTRODUCTION
Generator excitation system maintains generator voltage
and controls the reactive power flow using an automatic
voltage regulator (AVR) [1]. The role of the AVR system is to
hold the terminal voltage magnitude of a synchronous
generator at a specified level. Hence, the stability of this
regulator is of great concern because it can seriously affect the
security of the power system.
In addition, due to the uncertainties and nonlinearities of
some industrial plants, tuning PID parameters becomes
difficult. Several heuristic and metaheuristic tuning methods
have been considered, among which PSO [2] and GA [3].
They are based on seeking the optimum of cost functions
which well code the desired performances of the AVR.
However selection of the appropriate cost function to be
optimized as well as the best parameters of the optimizer is not
always an easy task. In this paper, we propose to implement a
metaheuristic particle swarm optimization (PSO) with
different topologies to optimize different cost functions in
sake of appropriately tune PID gains to improve AVR system
performance.
Next section describes the AVR system model. Section III
introduces PSO algorithm and some of its basic concepts that
are necessary to the understanding of this work. Section IV
describes the conducted experiments and simulation results
and analysis. Then, section V concludes this paper.
2015 IEEE
0.2497
Ts (s)
2%
0.3722
Po
(%)
1.9922
0.1385
4.1790
4.0565
0.0024
F3
0.0797
0.8388
33.9008
0.0053
F4
0.0797
0.8388
33.9008
0.0053
F5
0.2046
0.5553
6.7463
0.0048
F6
0.2638
0.3963
1.4653
0.0016
F7
0.2698
0.4051
1.4829
0.0065
F8
0.25
0.3728
1.9812
0.0043
F9
0.2497
0.3722
1.9922
0.0043
Name
Tr (s)
F1
F2
ess
0.0043
k: vector [kpkikd]
Po: percent overshoot
Ts: settling time
Tr: rise time
Tp: peak time
ess: steady state error
bw: bandwidth
mse: mean square error
ise: integral square error
itse: integral time square error
iae: integral absolute error
itae: integral time absolute error
set
Step Response
1.4
tf
F( Gbest)
Tr (s)
Ts (s)
2%
Po
(%)
ess
R
I
N
G
CP
426
5724.3
0.2501
0.3731
1.9738
0.0043
LP
501
5704.6
0.2495
0.3720
1.9995
0.0043
EP
368
5741.1
0.2497
0.3722
1.9926
0.0043
F
U
L
L
CP
157
5523.3
0.2499
0.3727
1.9843
0.0043
LP
224
5705.2
0.2496
0.3721
1.999
0.0043
EP
153
5724.2
0.2499
0.3726
1.9941
0.0042
S
T
A
R
CP
188
5706.6
0.2497
0.3723
1.9987
0.0042
LP
320
5704.8
0.2495
0.3720
1.9979
0.0043
EP
121
5746.4
0.2523
0.3766
1.9728
0.0030
1.2
Amplitude
TABLE IV.
performance.
Tr (s)
PO (%)
0.3727
0.2499
1.9843
0.5155
0.3433
0.5200
0.2400
2%
0.4
0.2
PSO
PSO
[7]
PSO
[8]
ess
0.5
1.5
2.5
Time (seconds)
V. CONCLUSION
This paper examined the optimization of a PID controlled
AVR system using different topologies of PSO and also
different cost functions. Comparison of obtained PID
controllers has been performed and encouraging system
performance has been obtained. Comparison of this
performance with reference reported results has been included.
Beyond suggesting different fitness functions, we have
designed a different formula that well describe the problem and
inspires a different search approache. In addition, one of the
explored coefficient profiles, exponential variation, effectively
enhances speed of convergence. Finally, this work proved that
despite the considerable progress recorded using PSO
algorithm, much is still to be achieved as relating time
performance to frequency domain characteristics, and
establishing appropriate coefficient control over the swarm
search behavior.
REFERENCES
TABLE STYLES
Kp
Ki
Kd
43
0.6835
0.6322
0.2722
1.0195
NM*
0.5857
0.4189
0.1772
2.6000
NM*
0.7080
0.6560
0.2820
(*10-4)
[7]
Ts (s)
Our AVR
0.6
[8]
TABLE V.
0.8
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
3.5