You are on page 1of 15

Online Voltage Stability Assessment of Power System

by Comparing Voltage Stability Indices and Extreme


Learning Machine

M.V. Suganyadevi and C.K. Babulal

Department of Electrical and Electronics Engineering,


Thiagarajar College of Engineering, India
{mvseee,ckbeee}@tce.edu

Abstract. Nowadays the modern power systems are large complex systems and
widely distributed geographically. The increase in demand, generator reaching
reactive power limits and line/generator outages may operate the power system
in stressed conditions leads to voltage instability or voltage collapse. Voltage
stability has become of major concern among the power utilities, because of
several events of voltage collapse occurred in the past decade. Thus the real
time voltage stability assessment is essential by estimating the loadability
margin of the power system. Voltage stability analysis was studied on IEEE 30
and IEEE 118 bus systems using many voltage stability indices based on
line/nodal and the results are compared with the proposed new index based on
SVM and ELM.

Keywords: voltage stability assessment, voltage stability index, support vector


machine, extreme learning machine.

1 Introduction

Problems related to voltage stability have recently been considered as the major
concerns in the planning and operation of power systems. The rapid increase in load
demand in electric power system motivates the researchers to protect the power
system to restrain voltage collapse. Voltage stability is concerned with the ability of
power systems to maintain acceptable voltages at all buses in the system under normal
conditions and after being subjected to a disturbance[1-3]. Voltage instability has
been attributed to the lack of adequate reactive support and the difficulty in the flow
of required reactive power on the transmission network.
In most of the cases, voltage profiles show no abnormality prior to undergoing
voltage collapse because of the load variation. Voltage stability margin (VSM) is a
static voltage stability index which is used to quantify how “close” a particular
operating point is to the point of voltage collapse [4]. Thus, VSM may be used to
estimate the steady-state voltage stability limit of a power system. Knowledge of the

B.K. Panigrahi et al. (Eds.): SEMCCO 2013, Part I, LNCS 8297, pp. 710–724, 2013.
© Springer International Publishing Switzerland 2013
Online Voltage Stability Assessment of Power System 711

voltage stability margin is of vital importance to utilities in order to operate their


system with appropriate security and reliability. The system operator must be
provided with an accurate and fast method to predict the voltage stability margin so as
to initiate the necessary control actions.
During the last few years, several methodologies for detecting the voltage collapse
points (saddle-node bifurcations) in power systems using steady-state analysis
techniques have been modified and applied for the determination of analyzing voltage
stability of power systems for example PV and QV curves, sensitivity-based indices
and continuation power flow methods [5,6]. Other methods, such as bifurcation
theory[7], energy function, singular value decomposition and so forth, have been also
reported in the literature.
Several static voltage stability methods are also available in the literature for the
quick assessment of voltage stability. In literature [8-16] several methods have been
proposed to identify critical bus bars, critical line, and stability margins of the power
system.
Some of the indices are:

1. Line Stability Index (Lmn).


2. Fast Voltage Stability Index (FVSI).
3. Line Stability Factor (LQP).
4. Voltage Collapse Proximity Indices (VCPI).
5. L-Index (L).
6. Three Diagonal Element dependent Index (Ipi, Iqi, Ii).

This paper compares the performance of the above said voltage stability indices for
different loading scenarios like real power load increase, reactive power load increase
and both real and reactive power load increase of IEEE 30 and IEEE 118 bus test
systems [17]. In addition a new index is proposed based on Extreme Learning
Machine (ELM). The proposed index takes real and reactive power load as input
parameters and gives a voltage stability margin called ELM-VSI.
The Extreme learning machine (ELM) was proposed recently
as an efficient learning algorithm for single-hidden layer feed forward neural
network (SLFN) [19]. The Extreme learning machine (ELM) [20-23] studies a much
wider type of “generalized” SLFNs whose hidden layer need not be tuned. ELM was
originally developed for the single-hidden-layer feed forward neural networks and
then extended to the “generalized” SLFNs which may not be neuron alike and
compared with the index based on SVM [18].
Organization of the paper is as follows: Section 2 describes various VSI used to
examine the voltage stability of the system. Section 3 presents the ELM and the test
systems and analysis tools used are presented in section 4. Results and discussions are
presented in section 5. Finally conclusions are summarized in section 6.
712 M.V. Suganyadevi and C.K. Babulal

2 Voltage Stability Margin

Voltage stability margin is defined as difference between maximum transferring


power that system can be tolerated and power of normal operation. In this paper, this
voltage stability margin is referred to as the loadability margin and used as
performance index for voltage stability analysis. The purpose of VSI is to determine
the point of voltage instability, the weakest bus in the system and the critical line
referred to a bus. Indices proposed based on bus are known as Nodal Indices and that
of transmission lines are Line Indices. The Nodal Indices and Line Indices are briefly
discussed in the following section.

2.1 Line Voltage Stability Indices


The line indices are based on the bus power (real and reactive), voltage magnitude,
phase angle and impedance of the transmission line as shown in the Fig.1. The factor
D=  r
2
  r2  for stable operating condition. If the factor D is
( x sin δ + cos δ )Vi  − 4 x + x Q j ≥ 0
   
less than zero, it indicates the roots are imaginary and the system is unstable.

Bus i Bus j

Vi ∠δi V j ∠δ j
Pi,Qi r+jx Pj, Qj

Fig. 1. Typical one line diagram of transmission line

The Line Stability index Lmn derived from the factor D assuming lossless
transmission line is as follows
4 XQj
Lmn =
[Vi sin(θ − δ )]2 (1)
Line stability factor LQP index is based on a concept of power flow through a
single line. LQP is calculated as
 X  X  (2)
LQP = 4 2  2 P i 2 + Q j 
  V i 
 Vi 
The fast voltage stability index FVSI is formulated based on a power transmission
line.

4Z 2Q j
FVSI = 2
(3)
Vi X
Online Voltage Stability Assessment of Power System 713

The Voltage Collapse Proximity Index VCPI investigates the stability of each line
of the system and they are based on the concept of maximum power transferred
through a line.
Pj (4)
VCPI ( p ) =
Pj (max)

Qj (5)
VCPI ( Q ) =
Q j (max)

Where the values Pj and Qj are obtained from conventional power flows
calculations, and Pj(max) and Qj(max) are the maximum active and reactive power that
can be transferred through the line. The line indices varies from 0(no load) to 1
(maximum loadability). The value of indices that is evaluated close to 1.00 indicates
that the particular line is closed to its instability that may lead to voltage collapse. To
maintain system security the value of indices should be maintained well below the
value of 1.00.

2.2 Nodal Indices


The L-Index is a quantitative measure for the estimation of the distance from the
actual state of the system to the stability limit. The index can be computed as
 i∈nG F jiVi
L = max (L j ) = max 1 − (6)
j∈nL j∈n Vj

Fji is the submatrice of [H]


 Z LL FLG 
Where H = 
 K GL YGG 

Lj is a local indicator that determinates the busbars from where collapse may
originate. The L index varies in a range between 0 (no load) and 1 (maximum
loadability).

Diagonal element dependent indices I, Ip, Iq


These indices are observed that with increase in load at the load bus, the value of
diagonal elements of the Jacobian ∂Qi / ∂Vi and ∂Pi / ∂δ i gets reduced. Thus,
the deviation in value of ∂Qi / ∂Vi and ∂Pi / ∂δ i from its no-load value to the
value at any particular loading condition can be used as index of voltage stability for
the load bus i. Using this criteria two voltage stability indices are proposed:
∂Qi / ∂Vi (7)
I qi =
− Bii
714 M.V. Suganyadevi and C.K. Babulal

∂Pi / ∂δ i (8)
I pi =
− Bii

The threshold value of the VSI are proposed as

∂Pi / ∂δ i
Ii = N
(9)
B V
j =1
ii j

j ≠i
th
Where Bii is the susceptance of the line connected to i load bus

3 Extreme Learning Machine


The ELM algorithm was originally proposed by Huang et al. in [20] and it makes use
of the SLFN. The main concept behind the ELM lies in the random initialization of
the SLFN weights and biases
f(x) = h(x) (10)
where h(x) is the hidden-layer output corresponding to the input sample x and is
the output weight vector between the hidden layer and the output layer. One of the
salient features of ELM is that the hidden layer need not be tuned. Essentially, ELM
originally proposes to apply random computational nodes in the hidden layer, which
are independent of the training data. Different from traditional learning algorithms for
a neural type of SLFNs [24], ELM aims to reach not only the smallest training error
but also the smallest norm of output weights. ELM [25,26] and its variants [27,28]
mainly focus on the regression applications. Latest development of ELM has shown
some relationships between ELM and SVM [29,30].

4 Test System and Analytical Tool

For the analysis of voltage stability, the test systems – IEEE 30 and IEEE 118 bus
systems are considered. All the results are produced with the help of a program
developed in PSAT. PSAT is MATLAB software for electric power system analysis
and control [32]. Though several loading pattern are tested for two test systems, due
to space limitations only some scenarios are presented. In scenario 1, real power
load at the weakest load bus alone increased. In scenario 2, reactive power load at
the weakest load bus alone increased. In scenario 3, both real and reactive power
load at the all load buses are increased simultaneously. The above procedure is
illustrated in Fig. 3
Online Voltage Stability Assessment of Power System 715

Start

Run CPF Program using


PSAT for the given test
System

Iteration=1

Obtain Voltage, Phase


Angle, Real and Reactive
Power

i=1

Compute Lmn, LQP,


FVSI, LVSI, VCPI (p), VCPI
(Q)

NO

Last line
i=l+1
done

Yes

YES
Index<
Iteration=iteration+1
1
NO

Obtain Loadability limit


corresponding to VSI=1

End

Fig. 2. Flow Chart for the estimation of loadability margin using VSI
716 M.V. Suganyadevi and C.K. Babulal

5 Results and Discussions

5.1 Estimation of Loading Margin Using VSI and CPF of IEEE 30 Bus System

In order to compare the VSI, the loadability margin of IEEE 30 bus system is
calculated for three scenarios. Scenario 1, the real power demand at weakest load bus
is increased gradually. In scenario 2, the reactive power demand at the weakest load
bus is increased gradually. Finally in scenario 3, both the real and reactive power
demand at the weakest load bus is increased gradually.
In Scenario 1: The real power demand at bus 24 is 10% increased from its base
value. Continuation Power Flow (CPF) available in PSAT is used to trace the PV
curve for the above case. The real and reactive power load obtained in each iteration
of the CPF is used to calculate the voltage stability indices (Lmn, LQP, FVSI, VCPI (p),
VCPI (q), L, I, Ip, and Iq). Fig.4. (a) to (d) shows the variation of line voltage stability
indices and Fig.4.(e) to (g) shows the variation of nodal voltage stability indices with
respect to real power demand at bus 24. The following observations are made from
the figures 4. (a) to (d).

Line Voltage Stability Indices:


• Maximum loadability limit at bus 24 obtained from CPF is Pmax = 0.055 p.u.
• As real power load increases the FVSI, Lmn, LQP are also increases as
expected.
• However FVSI, Lmn, LQP are unable to detect the Pmax. Because FVSI, Lmn,
LQP are less than ‘1(p.u)’ at Pmax. This clearly indicates that FVSI, Lmn, and LQP gives
higher loadability limit than the actual value from CPF.
• VCPI (p) increases gradually to Pmax. Hence, VCPI (p), exactly determines the
loadability limit for this scenario.

Nodal Voltage Stability Indices:


Inspection of Fig. 4 (e) to (g reveals the following observation.
• As real power load increases, L, ‘I’ and Ip indices are also increases.
• But the L index is unable to detect the Pmax exactly. Because it gives a value
greater than 1.00 value at Pmax.
• The ‘I’ and Ip indices threshold value is 0.5 at the maximum loading point.
This clearly indicates that ‘I’ and Ip indices gives higher loadability limit than the
actual value from CPF.
In Scenario 2 : The reactive power at bus 24 is 10% increased from its base case
value. Continuation Power Flow available in PSAT is used to trace the QV curve. For
the same real and reactive power load obtained in each iteration of the CPF, the
voltage stability indices (Lmn, LQP, FVSI, VCPI (p), VCPI (q), L, I, Ip, Iq) are
determined. The results are compared in the following table. Qmax is the maximum
loadability limit at bus 24. First row in Table 1 presents the value of VSI
corresponding to maximum loadability limit and the Second row indicate the value of
loadability limit corresponding to each index at its threshold.
Online Voltage Stability Assessment of Power System 717

Line Voltage Stability Indices:


As reactive load power increases the FVSI, Lmn, LQP are also increases as expected.
However FVSI, Lmn, LQP are unable to detect the Qmax. Because FVSI, Lmn, LQP are less
than 1 at Qmax. This clearly indicates that FVSI, Lmn, LQP give higher loadability limit
than the actual. VCPI (VCPI (p) & VCPI (q)) increases gradually at Qmax, VCPI becomes
1.Hence VCPI, exactly determines the loadability limit for this load scenario 2.

Table 1. Loadability margin from VSI and CPF for scenario 2 and scenario 3
Load at 24th bus
Scenario 2 LM=0.067 (p.u) Scenario 3 LM= 0.095 (p.u)
VSI CPF VSI CPF
LMN (P.U) 1.5894 0.5876 1.009 0.8790
LQP (P.U) 1.0780 0.7971 1.089 0.7792
FVSI (P.U) 0.9923 0.4002 1.047 0.6602
VCPI(P) (P.U) 1.0001 0.9534 0.9323 0.9156
VCPI(Q) (P.U) 0.9899 0.9221 0.9329 0.9166
L (P.U) 2.5673 0.8794 1.304 0.8999
IP (P.U) 0.6932 0.3602 0.7930 0.4537
IQ (P.U) 0.6934 0.3554 0.7291 0.3777
I (P.U) 0.4534 0.1955 0.6665 0.3471

Nodal Voltage Stability Indices:


Generally L index increases with the reactive load increase. But it is unable to detect the
Qmax exactly. Because L index gives a greater value than the actual load at Qmax. The ‘I’, Iq
and Ip indices are also increase with the reactive load. Its threshold value is 0.5 at the
maximum loading point. So the indices values are always greater than its actual Qmax value.

Fig. 3. (From clockwise) (a) Real power Vs voltage and FVSI, (b) Real power Vs voltage and
LQP, (c) Real power Vs voltage and VCPI, (d) Real power Vs voltage and Lmn, (e) Real power
Vs , voltage and L, (f) Real power Vs voltage and Ip, (g) Real power Vs voltage and I
718 M.V. Suganyadevi and C.K. Babulal

In Scenario 3: In this scenario, both real and reactive power load is incremented
by 10%. In this type of loading also, LQP, Lmn, FVSI, I, Ip, Iq indices unable to detect
the maximum loadability limit exactly except VCPI (VCPI (p) & VCPI (q)) index. The
results are presented in Table 1.

5.2 Estimation of Loading Margin Using VSI and CPF of IEEE 118 Bus System
The loadability margin of IEEE 118 bus system is calculated for all the three
scenarios namely increase in the real power demand, reactive power demand and
both real and reactive power.
In Scenario 1, The real power demand at bus 88 is 10% increased from its base
value. Continuation Power Flow (CPF) available in PSAT is used to trace the PV
curve for the above case. The real and reactive power load obtained in each iteration
of the CPF is used to calculate the voltage stability indices (Lmn, LQP, FVSI, VCPI (p),
VCPI (q), L, I, Ip, and Iq). From the Table 2, the following observations are made.
• The maximum loadability limit at bus 88 obtained from CPF is Pmax = 2.05 p.u.
• As real power load increases the Line Voltage Stability Indices FVSI, Lmn,
LQP are also increases as expected.
• However FVSI, Lmn, LQP are unable to detect the Pmax. Because FVSI, Lmn,
LQP are less than ‘1’ at Pmax. This clearly indicates that FVSI, Lmn, and LQP gives
higher loadability limit than the actual value from CPF.
• VCPI (p) increases gradually to Pmax. Hence, VCPI (p) exactly determines the
loadability limit for this scenario 1.
• As real power load increases, Nodal Voltage Stability Indices L, ‘I’, Iq and Ip
indices are also increases.
• But the L index is unable to detect the Pmax exactly. Because L index gives a
value greater than 1.00 value at Pmax.
• The ‘I’, Iq and Ip indices threshold value is 0.5 at the maximum loading point.
This clearly indicates that the indices give higher loadability limit than the actual
value from CPF.

In Scenario 2
The reactive power at bus 88 is 10% increased from its base case value.
Continuation Power Flow available in PSAT is used to trace the QV curve. For the
same real and reactive power load obtained in each iteration of the CPF, the voltage
stability indices (Lmn, LQP, FVSI, VCPI (p), VCPI (q), L, I, Ip, Iq) are determined. The
results are compared in the following table. Qmax is the maximum loadability limit at
bus 24. From the table results, As reactive load power increases the Line VSI FVSI,
Lmn, LQP are also increases as expected. However FVSI, Lmn, LQP are unable to
detect the Qmax. Because FVSI, Lmn, LQP are less than 1 at Qmax. This clearly
indicates that FVSI, Lmn, LQP give higher loadability limit than the actual. VCPI
(VCPI (p) & VCPI (q)) increases gradually at Qmax, VCPI becomes 1.Hence VCPI,
exactly determines the loadability limit for this load scenario 2. And the Nodal
Voltage
Online Voltage Stability Assessment of Power System 719

Stability Indices L,’I’, Iq, Ip indices also increases with the reactive load increase. But
the indices are unable to detect the Qmax exactly. Because the indices give a greater
value than the actual maximum loadability limit Qmax. ‘I’, Iq and Ip index’s threshold
value is 0.5 at the maximum loading point. So the indices values are greater than its
actual maximum loadability margin.
In Scenario 3, both real and reactive power load is incremented by 10%. In this
type of loading also, LQP, Lmn, FVSI, I, Ip, Iq indices unable to detect the maximum
loadability limit exactly except VCPI (VCPI (p) & VCPI (q)) index. The results are
presented in the following Table 2 .

Table 2. Loadability Margin from VSI and CPF of IEEE 118 Bus Systems
Load at 88th bus
Scenario 1 Scenario 2 LM=2.99 (P.U) Scenario 3 LM=3.04 (P.U)
LM=2.05 (P.U)
VSI CPF VSI CPF
VSI CPF
LMN (P.U) 1.009 0.8790 1.138 0.879
1.400 0.886
LQP (P.U) 1.089 0.7792 1.084 0.778
1.086 0.945
FVSI (P.U) 1.047 0.6602 1.146 0.854
1.289 0.876
VCPI(P) (P.U) 0.9323 0.9156 1.118 0.998
1.112 0.995
VCPI(Q) (P.U) 0.9329 0.9166 1.115 0.993
1.114 0.996
L (P.U) 1.304 0.8999 1.096 0.755
1.776 0.453
IP (P.U) 0.7930 0.4537 0.559 0.344
0.999 0.339
IQ (P.U) 0.7291 0.3777 0.641 0.398
0.997 0.337
I (P.U) 0.6665 0.3471 0.663 0.345
0.664 0.442

5.3 Estimation of Loading Margin by SVM and ELM


Generation of Data
The required data is generated using CPF method available in PSAT. In IEEE 30 bus
test system, As many as 597 patterns were generated by changing the load at each bus
and generation randomly in wide range (± 50% of base case). Thus 35820 (597x60)
load samples were generated. Power factor at all load buses are maintained constant.
Out of 35820 data samples in IEEE 30 bus system, 80% of total samples (28656)
were selected arbitrarily for training, while 20% (7184) were used for testing. The
data samples used for testing the SVM model and ELM model are unseen values that
are not used in training. In IEEE 118 Bus Test System, As many as 300 patterns were
generated by changing the demand and generation at each bus randomly in wide range
(± 50% of base case). Totally 35400 data samples were generated. Out of 300
patterns, 80% load scenarios (28320 samples) were arbitrarily selected for training
while, 20% load scenarios (7080 samples) were used for testing the performance of
the estimation of loadability margin.
720 M.V. Suganyadevi and C.K. Babulal

Algorithm

Power System Network

Data Generation
Run the CPF using PSAT

Create a data base for the input vector in the form of real and reactive power load. The
target or output vector is in the form of lambda (loading margin).

Regression Scheme

SVM Regression ELM Regression


Select C and γ parameter, kernel select hidden node neurons, activation
t f ti

Loadability Margin

Fig. 4. Flow Chart for the estimation of LM using SVM and ELM

Table 3. LM comparisons among ELM, CPF, and SVM


IEEE 30 bus system IEEE 118 bus system
No
CPF SVM ELM CPF SVM ELM
1 1.18906 1.0148 1.01484 1.066508 1.07684 1.076811
2 1.00116 1.3205 1.32049 1.168095 1.06470 1.064689
3 1.60025 1.6073 1.60731 1.264029 1.24729 1.247270
4 2.14578 1.8738 1.87386 1.354573 1.32478 1.324782
5 2.09961 2.1188 2.11875 1.439952 1.42737 1.427375
6 2.47358 2.3406 2.34065 1.906011 1.89737 1.897375
7 2.86492 2.5376 2.537659 1.954922 1.96516 1.965166
8 2.57753 2.7067 2.706636 1.999516 2.02815 2.028154
9 3.05481 2.8395 2.839433 2.039769 2.08643 2.086434
10 3.01280 2.8809 2.880869 2.075612 2.14008 2.140087

The voltage stability margin by CPF, SVM model and ELM are compared in
Table 3 respectively for few testing patterns due to limited space. But the CPF is
unable to detect the loading margin exactly, because it gives a value greater than 1.00
value at Pmax when compared to ELM and SVM. The tables show clearly that the
proposed ELM model estimate the same loadability margin as obtained by the
conventional techniques with greater accuracy. The MSE values of IEEE test systems
were simulated for SVM and ELM for two different activation functions namely
sigmoid and RBF are tabulated in Table 4. The training computational time of
Online Voltage Stability Assessment of Power System 721

SVM-sigmoid is slightly higher than the SVM-RBF. The ELM-RBF the training and
testing time are faster and accurate when compared to its sigmoid activation function.
The SVM and ELM MSE are also tabulated in order to show the accuracy of the
regression types. However the results show that the RBF types predicts the result
quickly when compared to sigmoid type and the computational time also less for the
estimation of loadability margin of a power system. The results show that the ELM
-3
network is able to produce the output with good accuracy(10 ). The MSE and the
-4
computational time for the system obtained are also very less in the order of 10 and
in few seconds respectively.

Table 4. Comparison of SVM and ELM

Regression Training data Testing data Activation Training Testing MSE


Scheme samples samples Function Time Time
(sec) (sec)
-005
IEEE 30 Bus SVM 28656 7184 Sigmoid 9.812 0.0998 3.255e
-006
System RBF 7.956 0.0447 2.192e
-005
ELM 28656 7184 Sigmoid 7.462 0.0863 1.881e
-006
RBF 5.578 0.0274 1.005e
IEEE 118 Bus SVM 28320 7080 Sigmoid 25.665 0.134 3.234e-005
-006
System RBF 18.830 0.116 2.784e
ELM 28320 7080 Sigmoid 23.436 0.129 1.925e-005
-006
RBF 16.991 0.109 1.028e

The Fig.6 and Fig 7 shows the prediction of loadability margin for IEEE 30 and
IEEE 118 test bus systems respectively by comparing the two regression scheme
models: SVM and ELM in terms of MSE for some testing patterns. The ELM-RBF
predicts the loadability margin more quickly and accurately when compared to
SVM-RBF.
-4
x 10
3
SVM-RBF
ELM-RBF
2.5

1.5
MSE

0.5

-0.5
0 20 40 60 80 100 120
Testing Patterns

Fig. 5. Comparison of MSE among SVM and ELM of IEEE 30 bus system
722 M.V. Suganyadevi and C.K. Babulal

-4
x 10
3
SVM-RBF
ELM-RBF
2.5

1.5
MSE

0.5

-0.5
0 20 40 60 80 100 120
Testing Patterns

Fig. 6. Comparison of MSE among SVM and ELM of IEEE 118 bus system

6 Conclusion
In this paper, a new voltage stability index based on extreme learning machine
technique is proposed which is used to evaluate the power system voltage stability.
The performance of proposed ELM-VSI and all Voltage stability indices (Lmn, LQP,
FVSI, VCPI (p), VCPI (q), L, I, Ip, Iq) has been tested on IEEE 30 bus and IEEE 118 bus
systems to evaluate the loadability margin of power systems. The already proposed
voltage stability indices (Lmn, LQP, FVSI, VCPI (p), VCPI (q), L, I, Ip,IQ) in the literature
are unable to predict the exact loadability margin (LM) as that of CPF. The VSI based
on ELM technique could able to identify the expected LM value with an error
-4
occurrence of 10 . A comparison of the propose index with other indices indicates
that the ELM-VSI is a much more reliable indicator of the relative closeness to the
voltage stability limit of a heavily loaded power systems. The proposed ELM-VSI can
be implemented for on-line security assessment in Energy Management System

Acknowledgement. The first author sincerely acknowledges the financial assistance


received from Department of Science and Technology, New Delhi, India under
Women Scientist Scheme-A vide letter number SR/WOS-A/ET-139/2011, dated 05-
03-2012 and the authors are sincerely thanks the Management and Principal of
Thiagarajar College of Engineering, Madurai, India to carry out this research work.

References
[1] Voltage Stability of Power Systems: Concepts, Analytical Tools and Industry
Experience, IEEE Committee, vol. IEEE/PES 93TH0358-2- PWR (1990)
[2] Canizares, C.A. (ed.): Voltage Stability Assessment: Concepts, practices and tools.
IEEE/PES Power System Stability Subcommittee Special Publication, Final Document
(August 2002)
[3] Kundur, P.: Power System Stability and Control. McGraw-Hill, New York (1994)
Online Voltage Stability Assessment of Power System 723

[4] Srivastava, L., Singh, S.N., Sharma, J.: Estimation of loadability margin using parallel
self-organizing hierarchical neural network. Computers and Electrical Engg. 26(2),
151–167 (2000)
[5] Ajjarapu, V., Christy, C.: The continuation power flow: a tool for steady state voltage
stability analysis. IEEE Trans. Power Systems 7, 416–423 (1992)
[6] Canizares, C.A., Alvarado, F.L.: Point of collapse and continuation methods for large
AC/DC systems. IEEE Trans. Power systems 8, 1–8 (1993)
[7] Canizares, C.A.: On bifurcation, voltage collapse and load modeling. IEEE Trans.
Power Systems 10, 512–518 (1995)
[8] Sode-Yome, A., Mithulananthan, N., Lee, K.Y.: A maximum Loadability Margin
Method for static voltage stability in Power Systems. IEEE Transaction on Power
Systems 21, 799–808 (2006)
[9] Ioannis, K., Konstantinos, O.: An Analysis of Blackouts for Electric Power
Transmission Systems. Transmission on Engineering, Computing and Technology 12,
289–292 (2006)
[10] Kwatny, H.G., Fischl, R.F., Nwankpa, C.O.: Local Bifurcation in Power Systems:
Theory, computation, and applications. Proc. IEEE 83(11), 1456–1483 (1995)
[11] Bian, J., Rastgoufard, P.: Power System Voltage Stability and Security Assessment.
Electr. Power Syst. Res. 30(3), 197–200 (1994)
[12] Gubina, F., Strmcnik, B.: Voltage Collapse Proximity Index determination using
Voltage Phasors Approach. IEEE Transaction on Power System 10(2), 778–794 (1995)
[13] Moghavvemi, M., Omar, F.M.: Technique for Contingency Monitoring and Voltage
Collapse Prediction. IEEE Proceeding on Generation,Transmission and
Distribution 145(6), 634–640 (1998)
[14] Mohamed, A., Jasmon, G.B., Yusoff, S.: A Static Voltage Collapse Indicator using Line
Stability Factors. Journal of Industrial Technology 7(1), 73–85 (1989)
[15] Moghavvemi, M., Faruque, O.: Real-Time Contingency Evaluation and Ranking
Technique. IEEE Proceeding on Generation, Transmission and Distribution 145(5)
(September 1998)
[16] Chebbo, A.M., Irving, M.R., Sterling, M.J.H.: Voltage Collapse Proximity Indicator:
behaviour and implications. IEEE Proc.-C 139(3) (May 1992)
[17] Kessel, P., Glavitsch, H.: Estimating the Voltage Stability of a Power System. IEEE
Transactions on Power Delivery, vol.PWRD-1(3) (July 1986)
[18] Sinha, A.K., Hazarika, D.: A Comparative study of Voltage Stability Indices in a Power
System. Electrical Power and Energy System 22, 589–596 (2000)
[19] Suganyadevi, M.V., Babulal, C.K.: Estimating of Loadability Margin of a Power
System by comparing Voltage Stability Indices. In: Proceeding of IEEE International
Conference on Control, Automation, Communication And Energy Conservation, June
4-6 (2009)
[20] Suganyadevi, M.V., Babulal, C.K.: Prediction of Loadability Margin of a Power System
using Support Vector Machine. In: Proceeding of IEEE International Conference on
Energy Efficient Technologies for Sustainability (ICEETS 2013), April 10-12 (2013)
[21] Huang, G.-B., Zhu, Q.-Y., Siew, C.-K.: Extreme learning machine: A new learning
scheme of feedforward neural networks. In: Proc. IJCNN, Budapest, Hungary, July 25-
29, vol. 2, pp. 985–990 (2004)
[22] Huang, G.-B., Zhu, Q.-Y., Siew, C.-K.: Extreme learning machine: Theory and
applications. Neurocomputing 70(1-3), 489–501 (2006)
724 M.V. Suganyadevi and C.K. Babulal

[23] Huang, G.-B., Chen, L., Siew, C.-K.: Universal approximation using incremental
constructive feedforward networks with random hidden nodes. IEEE Trans. Neural
Netw. 17(4), 879–892 (2006)
[24] Huang, G.-B., Chen, L.: Convex incremental extreme learning machine.
Neurocomputing 70(16-18), 3056–3062 (2007)
[25] Huang, G.-B., Chen, L.: Enhanced random search based incremental extreme learning
machine. Neurocomputing 71(16-18), 3460–3468 (2008)
[26] Smola, A.J., Scholkopf, B.: On a Kernel-Based Method for Pattern Recognition,
Regression, Approximation And Operator Inversion. Algorithmica 22, 211–231 (1998)
[27] Huang, G.B., Zhou, H., Ding, X., Zhang, R.: Extreme learning machine for regression
and multiclass classification. IEEE Transactions on Systems, Man, and Cybernetics,
Part B: Cybernetics, 1–17 (2010)
[28] Miche, Y., Sorjamaa, A., Bas, P., Simula, O., Jutten, C., Lendasse, A.: OP-ELM:
Optimally pruned extreme learning machine. IEEE Trans. Neural Netw. 21(1), 158–162
(2010)
[29] Tang, X., Han, M.: Partial Lanczos extreme learning machine for single-output
regression problems. Neurocomputing 72(13-15), 3066–3076 (2009)
[30] Liu, Q., He, Q., Shi, Z.-Z.: Extreme support vector machine classifier. In: Washio, T.,
Suzuki, E., Ting, K.M., Inokuchi, A. (eds.) PAKDD 2008. LNCS (LNAI), vol. 5012,
pp. 222–233. Springer, Heidelberg (2008)
[31] Frénay, B., Verleysen, M.: Using SVMs with randomised feature spaces: An extreme
learning approach. In: Proc. 18th ESANN, Bruges, Belgium, April 28-30, pp. 315–320
(2010)
[32] Power System Test Archive-UWEE (University of Washington),
http://www.ee.washington.edu/research/pstca
[33] PSAT, Power System Analysis Toolbox Version 2.1.1,
http://www.power.uwaterloo.ca/fmilano/downloads.html

You might also like