Professional Documents
Culture Documents
P.S.Venkataramu3
M.S.Nagaraja4
I.
INTRODUCTION
5
2
Optimal placement of SVC planning is necessary in order
to achieve enhancement of power system reactive power
(VAR) margin, reduction in system losses and voltage
depressions at critical points [12].
III.
QC 4
cos
1
XL
2
Series capacitor banks increases power transfer capability,
improve system stability, reduce system losses, improve the
voltage profile of lines, and optimize current sharing between
parallel lines. The reactive power supplied by the series
capacitor is calculated as equation 2 given below:
QC
V
K
XL 1 K
cos
ADALINE
The units with linear activation function are called linear units.
A network with a single linear unit is called an Adaline
(adaptive linear neuron) [13]. That is, in Adaline, the input
output relationship is linear. Adaline uses bipolar activation
for its input signals and its target output. The weights between
the input and the output are adjustable. The bias in Adaline
acts like an adjustable weight, whose connection is from a unit
with activations being always 1. Adaline is a net which has
only one output unit. The Adaline network may be trained
using delta rule. The delta rule may also be called as least
mean square (LMS) rule of Widrow Hoff rule. The learning
rule is found to minimize the mean squared error between the
activation and the target value [5].
A. The Adaline network training algorithm:
Step 0: Weights and bias are set to some random vales but not
zero. Set the learning rate parameter .
Step 1: Perform steps 2 6 when stopping condition is false.
Step 2: Perform steps 3 5 for each bipolar training pair s: t.
Step 3: Set activation for input units i=1 to n
x
s
Step 4: Calculate the net input to the output unit
The reactive power injected into the bus due to SVC is given
in equation 3 below.
.
3
For constant active power flow and supply voltage Vrms.
VAR (capacitive) = VAR (required) - VAR (uncompensated)
Capacitor susceptance
4
And required capacitor value is as
IV.
PROPOSED - ALGORITHM
In this section, numerical results are carried out on IEEE 30bus system shown in figure.3 to show the robust performance
and capabilities of ANN model [14].
Initially, the load flow solution using Newton - Raphson (NR)
method without contingency was carried out & voltages at
each bus were obtained. Next, for the same system the load
flow solution is obtained using NR method with contingency.
Table-I shows the post contingency voltage with the injected
MVAR. In some line outages there is no change in voltages
even after line outage and in some case there is complete black
out condition if line outage occurs as shown in Table-I.
TABLE I.
Line
Outage
(From To)
Critical
bus no
Base
case
Bus
voltage
Post
contingency
Bus voltage
MVAR
Injected
Bus
voltage
after
injection
1.022
0.931
98.532
1.022
30
0.995
0.974
3.1
0.995
30
0.995
0.985
1.193
30
0.995
0.974
2.99
0.995
30
0.995
0.974
2.961
0.995
30
0.995
0.983
1.735
0.995
30
0.995
0.979
2.209
0.995
30
0.995
0.986
0.481
0.995
1.033
0.946
17.536
1.008
30
0.995
0.994
0.141
0.995
30
0.995
0.99
0.62
0.995
10
11
10
12
12
13
12
14
30
0.995
0.994
0.134
0.995
12
15
30
0.995
0.989
0.85
0.995
12
16
30
0.995
0.993
0.167
0.995
14
15
30
0.995
0.994
0.03
0.995
16
17
30
0.995
0.994
0.072
0.995
15
18
30
0.995
0.994
0.016
0.995
18
19
30
0.995
0.994
0.002
0.995
19
20
30
0.995
0.994
0.023
0.995
10
20
20
1.029
0.985
7.594
1.029
10
17
10
21
26
1.001
0.993
0.695
0.995
10
22
30
0.995
0.993
0.215
0.995
21
22
15
23
30
0.995
0.989
0.734
0.995
22
24
30
0.995
0.987
0.973
0.995
23
24
30
0.995
0.992
0.296
0.995
24
25
30
0.995
0.988
0.796
0.995
25
26
25
27
26
1.001
0.982
1.909
1.001
28
27
30
0.995
0.861
10.521
0.995
27
29
29
1.006
0.949
3.993
1.006
27
30
30
0.995
0.94
4.383
0.995
29
30
30
0.995
0.977
1.875
0.995
28
30
0.995
0.993
0.155
0.995
28
30
0.995
0.97
2.428
0.996
NO CHANGE
COMPLETE BLACK OUT
30
0.995
0.991
0.466
0.995
NO CHANGE
COMPLETE BLACK OUT
NO CHANGE
NO CHANGE
TABLE II.
Line
Outage
(From To)
6
6
9
12
12
14
16
15
18
19
10
10
Input 2
Post
contingency
bus voltage
MVAR
injected
0.994
0.99
0.991
0.994
0.993
0.994
0.994
0.994
0.994
0.994
0.993
0.993
0.141
0.62
0.466
0.134
0.167
0.03
0.072
0.016
0.002
0.023
0.695
0.215
8
9
10
14
16
15
17
18
19
20
21
22
Target
Base
case
bus
voltage
0.995
0.995
0.995
0.995
0.995
0.995
0.995
0.995
0.995
0.995
1.001
0.995
Output
Bus voltage
Data after
training
0.9956
0.9972
0.9964
0.9956
0.995
0.9944
0.9949
0.9943
0.9941
0.9944
1.0007
0.9955
Errors
(mse)
TABLE V.
-0.0006
-0.0022
-0.0014
-0.0006
0
0.0006
0.0001
0.0007
0.0009
0.0006
0.0003
-0.0005
Input 1
Line Outage
(From - To)
Target
Output
Post
contingency
bus voltage
MVAR
injected
MVAR
predicted
Input 2
Errors
(mse)
Bus
voltage
after
MVAR
injection
23
24
0.992
0.296
0.2496
0.046
0.995
28
0.993
0.155
0.1666
-0.011
0.995
TABLE IV.
Input 2
Target
Post
contingency
bus voltage
MVAR
injected
Base
case
bus
voltage
0.985
0.983
0.986
0.989
0.989
0.987
1.193
1.735
0.481
0.85
0.734
0.973
0.995
0.995
0.995
0.995
0.995
0.995
Input 1
Line
Outage
(From To)
2
2
5
12
15
22
4
6
7
15
23
24
Line
Outage
(From To)
Input 1
Target
Output
Post
contingency
bus voltage
MVAR
injected
MVAR
predicted
24
25
0.988
0.796
0.8988
25
27
0.982
1.909
2.0036
TABLE VI.
Errors
(mse)
-0.0003
-0.0041
0.0061
-0.0004
0.0008
0.0001
0.1028
0.0946
Input 2
Target
Output
Post
contingency
bus voltage
MVAR
injected
Base
case
bus
voltage
Bus
voltage
Data
after
training
Input 1
Line
Outage
(From To)
Errors
(mse)
Input 2
Bus
voltage
after
MVAR
injection
0.995
1.001
Errors
(mse)
0.974
2.99
0.995
1.0032
-0.0082
0.974
2.961
0.995
1.0028
-0.0078
0.979
2.209
0.995
0.9972
-0.0022
29
30
0.977
1.875
0.995
0.9909
0.0041
Output
Bus
voltage
Data
after
training
0.9953
0.9991
0.9889
0.9954
0.9942
0.9949
weight is 0.8965 and for input 2 the weight is 0.0103. The bias
was 0.1 and error was 5.48e-05. No. of epochs were 72.
The testing results for the range shown in Table-IV are shown
in Table-V
Line
Outage
(From
- To)
1
6
3
28
Post
continge
ncy bus
voltage
0.974
0.97
Output
MVAR
injected
MVAR
predicted
3.1
2.428
2.3762
2.6407
Errors
(mse)
0.7238
-0.2127
Input 2
Bus
voltage
after
MVAR
injection
0.995
0.995
VI.
CONCLUSION
The line outage contingency create the under voltage and over
voltage condition in the system. The most critical bus was
identified by the voltage difference from the base case and the
contingency case. The most effected bus has been selected as
the point where MVAR has to be injected. The voltages has
been increased when MVAR was injected in the particular bus
so that the post contingency voltage becomes almost equal to
pre contingency voltage, thus the system becomes stable.
ANN technique was implemented for predicting the injected
MVAR at critical bus. Adaline was used for training & testing.
The network was trained by taking input 1 as post contingency
bus voltage, input 2 as MVAR injected and target as precontingency bus voltage. By this target, weights were obtained
for each input. These weights were used for testing the
network for obtaining the MVAR injected for unknown cases.
The predicted MVAR value was within 5% tolerance.
ACKNOWLEDGMENT
Authors are grateful to the management of Gyan Ganga
Institute of Technology &Management, Bhopal and also to the
management of Bapuji Institute of Engineering and
Technology, Davangere.
REFERENCES
[1]. Scott Greene, Ian Dobson, Fernando L. Alvarado Contingency Ranking
For Voltage Collapse Via Sensitivities From A Single Nose Curve ,
IEEE Transactions on Power Systems, Vol. 14, No. 1, February 1999.
[2]. G. Balamurgan and P. Aravindhababu, ANN based online estimation
of voltage collapse proximity indicator, International Journal of
Engineering Science and Technology Vol. 2(7), 2010,2869-2875.
[3]. Manish Jain, P.S. Venkataramu, T. Ananthapadmanabha Critical Bus
Ranking Under Line Outage Contingencies , Proceeding of ninth
Lasted International Conference Power & Energy Systems Jan35,2007,Clearwater, Florida, USA. ISBN Hardcopy: 978 0 88986
621 - 8 / CD / 978 0 88986 623 2
[4]. S. Tosun, A. ztrk, M. A. Yaln Investigation Of Critical Bus Values
In Electric Power System Using Simulated Annealing And Tabu Search
Algorithms , Scientific Research and Essays Vol. 5 (18), pp. 26732680,
18
September,
2010
Available
online
at
ISSN
1992-2248,
2010
http://www.academicjournals.org/SRE
Academic Journals.
[5]. S.N. Sivanandam and S.N. Deepa Principles of Soft Computing ,
Second Edition, 2011 Wiley India Private Limited
[6]. M. Tarafdar Haque, and A.M. Kashtiban Application of Neural
Networks in Power Systems; A Review.
[7]. Anjan Bose, Kevin Tomsovic, Liqiang Chen and Mohammad Vaziri,
Automated Operating Procedures for Transfer Limits Final Report,
Washington State University.
[8]. Yongan Deng, Reactive Power Compensation of Transmission Lines
MASc student at Concordia University.
[9]. A.H. Almasoud Effect of Series And Shunt Compensation of A Given
380 Kv Power System Network EE and CE Dept., King Abdulaziz
University, Jeddah, Saudi Arabia.
[10]. Omid Alizadeh Mousavi and Rachid Cherkaouri, Literature Survey on
Fundamental Issues of Voltage and Reactive Power Control, Swiss
Federal Institute of Technology Zurich.
[11]. Kundur. P. Power system stability and control, McGraw-Hill,
Inc.;1993.
[12]. S. Sakthivel, Dr. D. Mary, R. Vetrivel and V. Senthamarai Kannan,
Optimal Location of SVC for Voltage Stability Enhancement under
APPENDIX