World Academy of Science, Engineering and Technology 36 2009
An ACO Based Algorithm for Distribution
Networks Including Dispersed Generations
B. Bahmani Firouzi, T. Niknam, M. Nayeripour
Abstract—With Power system movement toward restructuring
along with factors such as life environment pollution, problems of
transmission expansion and with advancement in construction
technology of small generation units, it is expected that small units
like wind turbines, fuel cells, photovoltaic, … that most of the time
connect to the distribution networks play a very essential role in
electric power industry. With increase in developing usage of small
generation units, management of distribution networks should be
reviewed. The target of this paper is to present a new method for
optimal management of active and reactive power in distribution
networks with regard to costs pertaining to various types of dispersed
generations, capacitors and cost of electric energy achieved from
network.
In other words, in this method it’s endeavored to select optimal
sources of active and reactive power generation and controlling
equipments such as dispersed generations, capacitors, under load tapchanger transformers and substations in a way that firstly costs in
relation to them are minimized and secondly technical and physical
constraints are regarded. Because the optimal management of
distribution networks is an optimization problem with continuous and
discrete variables, the new evolutionary method based on Ant Colony
Algorithm has been applied. The simulation results of the method
tested on two cases containing 23 and 34 buses exist and will be
shown at later sections.
Keywords—Distributed Generation, Optimal Operation
Management of distribution networks, Ant Colony Optimization
(ACO).
I. INTRODUCTION
D
URING some last decades due to a great increase in
operation efficiency and encouragement of financiers,
electric power industry has encountered basic changes in
the light of management and ownership, in a way that for
making a proper competitive conditions, various parts such as
generation, transmission and distribution have been
independent from each other.
These changes along with factors like environment
pollution, transmission line establishment and technology
advancement in economical construction of small-scale
generation units in comparison with large ones have resulted
in an increase in the usage of small-scale ones under the topic
B. Bahmani Firouzi is with Islamic Azad University, Marvdasht Barnch,
Iran
T. Niknam and M. Nayeriopur are with Shiraz University of Technology,
Shiraz, Iran. E-mails : niknam@sutech.ac.ir and taher_nik@yahoo.com
named dispersed generations that mostly connect to
distribution networks without needing transmission lines.
Researches made by researching centers such as EPRI have
anticipated that until the year 2010, about 25 percent of
electric power is generated by dispersed generations.
Therefore with developing usage process of these generations,
field of management and operation should be studied more
carefully. Generally, optimal operation management of power
systems is applied to optimal usage of active and reactive
power generation equipments entirely and controlling devices.
The reason for is that firstly costs are minimized and secondly
technical and physical constraints are regarded.
In the past, distribution networks only consisted of reactive
power generation sources. Because of this, most of
explorations done in this part of power systems had to do with
optimal operation of reactive power[1-9]. But these days due
to existence of dispersed generations, the effects of various
types of these generations in the light of active and reactive
power generation should be considered.
This paper presents a method for optimal operation from
distribution networks with regard to cost effect of active and
reactive power generation consisting of dispersed generation
substations and capacitors in order that firstly cost of active
and reactive power generation and network losses are
minimized, secondly technical constraints are regarded too.
In other words, the object is to determine active and
reactive power generated by dispersed generations, main
substation (distribution offices), capacitors and also tapchanger transformers in a manner to minimize objective
function and regard the physical and technical constraints.
In overall view, because optimal operation of distribution
networks is an optimization problem including continuous and
discrete variables, evolutionary methods due to independence
on primary conditions, being differentiable and continuous
can be considered more and more.
One of evolutionary methods that have been considered
recently is implementation of finding shortest path process
done by ants. For the first time, Dorigo and his collaborators
proposed the usage of Ant Colony Method for solving
complicated optimization problems such as TSP (Traveling
Salesman Problem) and QAP (Quadratic Assignment
problem). Until now the Ant Colony Algorithm has been
applied for solving some optimization problems such as TSP,
ATSP, QAP, JSP, SMTTP, programming of Hydro electric
power generation, economic dispatch, unit commitment,
voltage and power control in distribution networks with
regard to the effects of dispersed generations and pricing
reactive power in restructured networks [10-18].
107
World Academy of Science, Engineering and Technology 36 2009
TABLE I
With help of new method based on Ant Colony Algorithm that
is presented in this paper, several optimization problems
consisting of continuous and discrete variables such as
operation management of distribution networks can be solved.
Then optimal operation management of distribution networks
with regard to the effects of dispersed generations along with
costs of electric power generation for various types of
dispersed generations are presented and after that, Ant Colony
Algorithm mechanism and its application to solve
optimization problems along with flowchart and solving
method are observed. Finally simulation results achieved
through the use of this Algorithm tested on two networks
containing 23 and 34 buses are shown.
COMPARISON OF SELECTED ELECTRICITY GENERATION
TECHNOLOGIES[20]
Micro turbine
Power Only
Micro
turbine-CHP
Gas ICEPower Only
Gas ICE-CHP
Fuel CellCHP
Solar
Photovoltaic
Small Wind
Turbine
Large Wind
Turbine
Combustion
TurbinePower Only
Combustion
Turbine-CHP
CombinedCycle System
II. OPTIMAL OPERATION MANAGEMENT OF DISTRIBUTION
NETWORKS WITH REGARD TO DISPERSED GENERATION
From a mathematical standpoint the optimal operation
management of distribution network with regard to distributed
generation is an optimization problem with inequality
constraints. The objective function is the summation of active
and reactive cost of DGs, reactive cost of capacitors and active
power cost of substation as follows:
N
N
N
N
f ( x) = C (P ) + C(P ) + C(Q ) + C(Q ) + Ploss * MCP (1)
∑
g
Sub
Sub
i =1
∑
g
gi
i =1
∑
c
gi
i =1
∑
b
ci
i =1
i
O&M
Cost
($/kWh)
Service
Life
(Years)
.075
.015
12.5
1765
.035
.015
12.5
100
1030
.067
.018
12.5
100
1491
.027
.018
12.5
200
3674
.029
.01
12.5
100
6675
0
.005
20
10
3866
0
.005
20
1000
1500
0
.005
20
100000
715
.067
.006
20
100000
921
.032
.006
20
1000000
690
.032
.006
20
Capacity
(kW)
Capital
Costa
($/kW)
100
1485
100
Fuel
Cost
($/kWh
Cost of DGs (per kWh/$), based on above table, can be
defined as follows:
where:
CSub is substation active cost.
C(Pg) and C(Qg) are active and reactive cost of DGs.
C(Qc) is reactive cost of capacitors.
Ploss is branch loss.
Nc is number of capacitors.
Ng is number of DGs.
Nb is number of branches.
MCP is market-clearing price.
C (P) = a + b * P
In mentioned equation a & b coefficients can be evaluated as(6)
follows:
a=
CapitalCos t ($ / kW ) * Capacity ( kW ) * Gr
LifeTime (Year ) * 365 * 24 * LF
(7)
Constraints are defined as follows:
• Active and reactive power constraints of DGs:
Pg min i < Pgi < Pg max i
Q g min i < Q gi < Q g max i
• Transmission line limits:
PLij < PL max ij
•
•
•
(2)
(3)
0 < Q ci < Q c max i
(4)
Tap min i < Tap i < Tap max i
(5)
b = FuelCost ($ / kWh ) + O & MCost ($ / kWh )
where Gr and LF are yearly rate of benefit and DG loading
factor.
The cost of reactive power produced by generators is called
opportunity cost which due to capability diagram of generator
shown in fig (1), reduces the active power production
capacity.
Reactive power of capacitors:
Tap of Transformers:
Load flow equations.
III. EVALUATION COST OF DISTRIBUTED GENERATION
Generally, costs of distributed generation to customers include
the installed cost of the equipment, fuel costs, nonfuel
operation and maintenance (O&M) expenses, and certain costs
that the customers’ utility imposes.
Table (I) shows comparison of different cost of some
distributed generations.
Fig. 1. Loading capability diagram
Opportunity cost depends on demand and supply in market,
so it is hard to determine its exact value. In simplest form
opportunity cost can be considered as follows:
2
2
(8)
Cgqi(QGi ) = Cgpi(SGi,max) −Cgpi SGi
,max − QGi .K
[
Where:
108
(
)]
World Academy of Science, Engineering and Technology 36 2009
S Gi , max : Maximum apparent power in ith bus
th
Q Gi : Reactive power of generator in i bus
K: Reactive power efficiency rate (usually between 5-10%)
IV. UNBALANCED THREE PHASE POWER FLOW
In unbalanced three-phase power flow, the following
components are modeled by their equivalent circuits in term of
inductance, capacitance, resistance and injected current.
a) Distributed Generators: DGs are modeled as constant
P and variable Q.
b) Transformers: transformers are modeled as
equivalent circuit with fictitious current injections.
c) Capacitors: Capacitors are represented by their
equivalent injected currents.
d) Demands or Loads: system loads are basically
considered asymmetrical; because of single-phase
loads and unequal three phase loads.
In this paper a network-topology-based on three-phase
distribution power flow algorithm is used. Two matrices are
used to obtain the power flow solution. They are the Bus
Injection to Branch Current (BIBC) and the Branch Current to
Bus Voltage (BCBV) matrices [19].
V. DISTRIBUTED GENERATION MODELING
Generally, depending on the contract and control status of a
generator, it may be operated in one of the following modes:
• To output power at a specific power factor (PQ node).
• To output power at a specific terminal voltage (PV
Node).
In general, DGs can be modeled four ways:
• PV model that each three phase can be controlled
instantaneously.
• PQ model that each three phase can be controlled
instantaneously.
• PV model that each phase could be controlled separately.
• PV model that each phase could be controlled separately.
We have used a reactive power compensation for modeling of
SVCs and PV nodes[9]. Fig2 shows model of DGs based on
kind of their control.
animals, they find the shortest path from nest to food with aid
of the pheromone. The pheromone is the chemical material
deposited by the ants, which serves as critical communication
media among ants, thereby guiding the determination of next
movement. On the other hand, ants find the shortest path,
based on intensity of pheromone deposited on different paths.
For better understanding, assume that ants want to move from
A to B and vice versa, to obtain food (Fig3).
Fig. 3. An example of finding the shortest path by ants
At first, if there is no obstacle, all of them will walk to the
straight path (Fig 3.a). Now, assume that there is an obstacle,
in this case, ants will not be able to follow the original trial in
their movement. Therefore, randomly, they turn to left (ACB)
and to right (ADB) (Fig 3.b). Since ADB path is shorter than
ACB, the intensity of pheromone deposited on ADB is more
than the other. So ants will be increasingly guided to move on
the shorter path (Fig 3.c). This behavior forms the
fundamental paradigm of ant colony system.
As it was indicated in Fig3, the intensity of deposited
pheromone is one of the most important factors for ants to find
the shortest path. Therefore, this factor should be used to
simulate behavior of ants. Generally, the following factors are
used to simulate ant systems:
• Intensity of pheromone
• Length of path
To select the next path, state transition probability is defined
as follows:
(τ ij ) γ (1 / L ij ) γ
(9)
P =
ij
∑
(τ ij ) (1 / L ij ) γ 2
1
2
γ1
After selecting the next path, trail intensity of pheromone is
updated as:
(10)
τ ij ( k + 1) = ρτ ij ( k ) + Δτ ij
Where:
τij :intensity of pheromone between nodes i and j, Lij: length of
path between nodes i and j,
ρ : a coefficient such that (1-ρ ) represents the evaporation of
trail between time k and k+1.
γ1 and γ2: control parameters for determining weight of trail
intensity and length of path.
Fig. 2. Model of DSs
a). PQ Model with instantaneously control
b). PQ Model with separately control
c). PV Model with instantaneously control
d). PV Model with separately control
VII. ANT COLONY ALGORITHM
VI. ANT COLONY SYSTEM MECHANISM
Ants are insects, which live together. Since they are blind
This section presents a new approach based on ant
algorithm for solving optimization problems. Optimization
109
World Academy of Science, Engineering and Technology 36 2009
problem is defined as:
Min
f(X)
s.t
hi(X ) = 0
i = 1,2,3,..., N eq
gi (X ) ≥ 0
(11)
i = 1,2,3,..., M
φ ij = F ( X i ) − F ( X j )
Transition probabilities are defined as:
(φ ij ) γ 1 (τ ij ) γ 2
Pij = K
( ∑ (φ ij ) γ 1 (τ ij ) γ 2 )
Where:
Neq: number of equality constraints,
M: number of inequality constraints,
X: state variables.
In order to apply ant colony algorithm the following steps
should be repeated.
Global _ Initial _ Colony _ Population = [ X 1 , X 2 ,..., X N ]
X i min ≤ X i ≤ X i max
Global _ Initial _ Intensity = [τ ij ] N * N
Local _ Trial _ Intensity = [τ ij ] M * M
Step 3: Determination of next path
Determination of next path for each colony of ants depends on
the direction of global and local paths. Namely, at first each
colony of ants has to find local and global path as follows:
1
j
* * 2
* *
*
*
*
*
*
* *
*
Value of K is equal to N and M for global and local transition
probabilities respectively.
The roulette wheel is used for stochastic selection. After
selection of local and global paths, trail intensity is updated as
follows:
Δ τ ij = P ij
(16)
τ ij ( k + 1 ) = ρτ ij ( k ) + Δ τ ij
Next path is determined based on local and global paths as
follows:
X i (k + 1) = X i (k ) + rand * ( X Local − X i (k )) + rand * ( X Global − X i (k )) (17)
Step 4: Check of convergence
After all of Ant colonies, find their next path, convergence is
checked by:
∑ (X
N
i =1
N
* *
*
1. Global path
direction
i
( k + 1) − X i ( k )) 2 < ε
(18)
If convergence condition is satisfied stop and print the results,
otherwise go to step 3.
VIII. FLOW CHART OF ALGORITHM
(13)
In this equation M is the number of ants in each colony and δ
is the radius of local area search.
i
*
(15)
New paths are compared with their limits.
(12)
where N is the number of Colonies.
Step 2: Creation of local initial population for each Ant
colony and local Trail Intensity
In this step for each ant colony, initial population is created
randomly. Also local trail intensity between ants in each
colony is generated.
X i − δ ≤ Yi ≤ X i + δ
(14)
j =1
Step 1: Creation of global initial population for Colonies and
Global Trail Intensity
An initial population of ant colonies, Xi that must meet
constraints, is selected randomly. At initialization phase it is
assumed that trail intensity between each two colonies are the
same
Local _ Initial _ Population = [Y1 , Y 2 ,..., Y M ]
two preceding directions (eq.9).
Selection of global and local path is based on (1). Since in
some optimization problems, Lij is not known, we can define
its inverse as follows:
Fig 5 shows flowchart of ant colony algorithm that
described in previous section.
The first step is to create an initial population (Global initial
population) for the colonies of ants based on control variables
(In this paper active and reactive power of DGs, reactive
power of capacitors and tap of LTC), which are between their
limits. Then an initial population (Local initial population)
will be created for each colony. In order to determine the next
path for each ant, global and local paths should be known.
Global and local paths determinations are similar. Using the
trail intensity, global and local transition probabilities are
calculated based on the difference between the cost of
colonies and the difference between the costs of ants in each
colony respectively. Afterward, global and local paths are
determined with roulette wheel. If convergence is met, it will
stop and otherwise the path determination steps are repeated.
Unbalanced three-phase power flow presented in [19] is used
to calculate the active power losses.
IX. SIMULATION
Fig. 4. Determination of next path for ant colony
The movement direction of any of ants is a combination of
In this section the proposed method is applied to optimal
operation management of distribution on two distribution test
110
World Academy of Science, Engineering and Technology 36 2009
feeders.
In following section results for two cases are presented. It
is assumed that energy price in substation is 4 cent per kWh
and Capital cost of capacitor banks can be considered as
deterioration rate and is written as follows:
C ci ( Q ci ) = Q ci × 11600 $
= Q ci × . 1324 $
M var
÷ ( H × 15 × 8760 ) hrs
in [21].
For this system it is assumed that there are three DGs
connected at 9, 23 and 27 respectively, which their
specification are presented in Table II.
(19)
M var .hr
Where H represents average duty cycle of capacitor banks and
value of 2 is considered for it in this study.
3
Read data include transmission line status,
DGs , Loads and Capacitors values
Fig. 6 Single Line Diagram
Create initial population and trial intensity
for Colony of ants based on DGs, Load and
Capacitors values
TABLE II
CHARACTERISTIC OF GENERATORS
G1
G2
Maximum Active Power(kW)
100
400
Maximum Reactive Power (Kvar)
80
320
Minimum Reactive Power (Kvar)
-60
-240
Location
6
16
Micro
Large
Kind of DG
Turbine
Wind
CHP
Turbine
Create initial population and local trial intensity for
each ant colony based on DGs, Load and Capacitors
values
G3
600
480
-360
29
Combustion
Turbine CHP
Table III give the comparison of results the proposed method
with Genetic Algorithm.
Calculate local transition
probabilities based on local
trial intensity and different
costs between ants in each
colony
Calculate global transition
probabilities based on global
trial intensity and different
costs between each colonies
Determination of local path
based on roulette wheel
Determination of global path
based on roulette wheel
Update local trail intensity
Update global trail intensity
TABLE III
COMPARISON RESULTS
Objective function Value ($/h)
Losses (Kw)
Tap of Substation Transformer
Tap of Voltage Regulator 1
Tap of Voltage Regulator 2
Active Power of DG1 (Kw)
Active Power of DG2 (Kw)
Active Power of DG3 (Kw)
Reactive Power of DG1 (Kvar)
Reactive Power of DG2 (Kvar)
Reactive Power of DG3 (Kvar)
Reactive Power of Capacitor 1(Kvar)
Reactive Power of Capacitor 2(Kvar)
Execution Time (S)
Calculate and determine next paths based on local and global paths
No
ACO
50.451
9.5528
1.0131
0.98
1.03
0
400
499.99
1.74
91.9974
95.81
450
0
300
GA
52.1954
17.5441
1.003
1.008
1.021
5.72
380.72
355.43
29.65
153.23
417.22
0
0
700
Case 2. A realistic 23 bus 20 Kv network
The method is applied to a rural network as shown in Figure
7. This system is used to supply power demand in the village
located in the north of Iran. Line and load characteristics are
shown in Tables IV and V respectively. Line impedance
Matrix is presented in equation (11). As there is no DG in this
networks currently, two typical DGs have been considered in
buses 13 and 21 which their specification have been presented
in Table VI. In this system there is one-capacitor (800Kvar),
which is located in bus 14.
Convergence condition is
satisfied
Yes
Stop and print results.
Fig. 5 Flowchart of proposed algorithm
Case 1: IEEE 34 bus radial test feeders
Figure 6 shows the IEEE 34 bus radial distribution test
feeders, where the lines and loads specification are presented
111
⎡ 7 + j 7 .2 + j.15 .2 + j.15 ⎤
Z Line (Ω / m ) = (1e − 4) ⎢⎢.2 + j.15 7 + j 7 .2 + j.15 ⎥⎥
⎢⎣.2 + j.15 .2 + j.15 7 + j 7 ⎥⎦
(20)
World Academy of Science, Engineering and Technology 36 2009
TABLE VI
CHARACTERISTIC OF GENERATORS
G1
Maximum Active Power
1000
Maximum Reactive Power
800
Minimum Reactive Power
-600
Location
13
Combustion Turbine
Kind of DG
CHP
Fig.7 Single Line Diagram of rural network
TABLE IV
LINE CHARACTERISTICS
No From To Length (m)
1
1
2
40
2
2
3
280
3
3
4
140
4
4
5
120
5
5
6
330
6
6
7
725
7
7
8
210
8
8
9
210
9
9
10
55
60
10 10 11
11 11 12
1000
12 12 13
1020
13 13 14
870
14 14 15
865
15 15 16
865
16 10 17
1400
17 17 18
1700
18 17 19
70
19 19 20
70
20 18 21
1060
21 21 22
1500
22 22 23
520
TABLE V
LOAD CHARACTERISTICS
No
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
Pa(Kw) Qa(Kvar) Pb(Kw) Qb(Kvar) Pc(Kw) Qc(Kvar)
0.00
0.00
0.00
0.00
0.00
0.00
105.00 78.75 114.45 85.84 95.55 71.66
83.33 62.50 90.83 68.13 75.83 56.88
83.33 62.50 90.83 68.13 75.83 56.88
83.33 62.50 90.83 68.13 75.83 56.88
83.33 62.50 90.83 68.13 75.83 56.88
83.33 62.50 90.83 68.13 75.83 56.88
83.33 62.50 90.83 68.13 75.83 56.88
83.33 62.50 90.83 68.13 75.83 56.88
83.33 62.50 90.83 68.13 75.83 56.88
105.00 78.75 114.45 85.84 95.55 71.66
105.00 78.75 114.45 85.84 95.55 71.66
83.33 62.50 90.83 68.13 75.83 56.88
83.33 62.50 90.83 68.13 75.83 56.88
21.00 15.75 22.89 17.17 19.11 14.33
333.33 250.00 363.33 272.50 303.33 227.50
133.33 100.00 145.33 109.00 121.33 91.00
83.33 62.50 90.83 68.13 75.83 56.88
105.00 78.75 114.45 85.84 95.55 71.66
105.00 78.75 114.45 85.84 95.55 71.66
50.00 37.50 54.50 40.88 45.50 34.13
0.00
0.00
0.00
0.00
0.00
0.00
105.00 78.75 114.45 85.84 95.55 71.66
G2
1000
800
-600
21
Combustion Turbine
CHP
A comparison between the proposed algorithm (ACO) and
Genetic Algorithm is available in Table VII.
TABLE VII
COMPARISON RESULTS
Objective function Value ($/h)
Losses (Kw)
Tap of Substation Transformer
Active Power of DG1 (Kw)
Active Power of DG2 (Kw)
Reactive Power of DG1 (Kvar)
Reactive Power of DG2 (Kvar)
Reactive Power of Capacitor (Kvar)
Execution Times (S)
ACO
254.0269
31.1441
1.03
764.3091
237.49
200.21
208.535
800
200
GA
254.0296
31.2657
1.0291
760.53
231.84
203.144
210.1121
800
423
As shown in Tables III and VII, the proposed method can
be used to apply to optimal operation management of
distribution networks. The results of these Tables can be
summarized as follows:
1. The execution time of proposed method is sufficiently
short (with regard to GA) and will give a general idea
that the method can be implemented without any
restriction in realistic networks.
2. The method can be applied to a wide variety of similar
optimization problems. On the other hand, this method
can be used to non-differential and non-continuous
objective function and constraints.
3. Objective function value and active power losses in the
proposed method is less than GA.
4. Because most of dispersed generations owned and
controlled by private sections, necessary mechanisms
must be applied for supervision and control of optimal
operation in power systems. In this paper costs
pertaining to active and reactive power generation
offered by owners of dispersed generations have been
used as a decisive factor for optimal control of them.
Results achieved in last sections show that we can apply
these methods to control dispersed generations and be
sure that high benefits will be gained from them.
X. CONCLUSION
As the number of DGs will be increasing, their impacts on
power system to be studied. One of the most important issues
in distribution system is distribution management system
(DMS), which can be affected by DGs. In this paper a new
approach for optimal operation management of distribution
networks with regard to DGs presented. The simulation result
showed that the method could be implemented in practical
distribution networks.
The execution time of proposed method is sufficiently short
and will give a general idea that the method can be
112
World Academy of Science, Engineering and Technology 36 2009
implemented without any restriction in realistic networks.
Since the most of DGs owned by private section, active and
reactive power generation costs of DGs considered as optimal
parameter control of them.
REFERENCES
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
M. E. Baran and M. Y. Hsu, “Volt/Var Control at Distribution
Substations”, IEEE Trans. On Power Systems, vol.14, No.1, Feb. 1999,
p.p. 312-318.
I.Roytelman, B.K. Wee and R.L.Lugtu, “Volt/Var Control Algorithm for
Modern Distribution Management System” IEEE trans. On Power
system, Vol. 10, No.3, Aug. 1995.
I.Roytelman and V. Ganesan, “Coordinated Local and Centralized
Control in Distribution Management Systems”, IEEE Trans. On Power
Delivery, vol.15, No.2, April 2000, p.p. 718-724.
V.Borozan, M.E. Baran and D. Novosel, “Integrated Volt/Var Control in
Distribution Systems” IEEE 2001, p.p.1485-1490.
Roytelman and V. Ganesan, “Modeling of Local Controllers in
Distribution Network Applications”, IEEE Trans. On Power Delivery,
vol. 15, No.4, Oct. 2000, p.p.1232-1237.
T.Niknam, A.M. Ranjbar and A.R. Shirani, “Impact of Distributed
Generation on Volt/Var Control in Distribution Networks”, IEEE
Bologna Power Tech 2003, Italy.
D.M. FalcZio, H.O. Henriques, “ Load estimation in radial distribution
systems using neural networks and fuzzy set techniques”, IEEE 2001,
P.P.1002-1006.
T.Niknam, A.M. Ranjbar and A.R. Shirani, “Volt/Var Control in
Distribution Networks with Distributed Generation”, IFAC Conference
2003, Korea.
T.Niknam, A.M. Ranjbar, and A.R. Shirani , “A New Approach Based
on Ant-Genetic Algorithm for Volt/Var Control in Distribution Network
with Considering Distributed Generation” IEEE DRPT2004 Hong Kong
April 2004.
M.Dorigo, G.D. Caro and L.M. Gambardella, “Ants Algorithms for
Discrete Optimization”, Artificial Life, Vlo.5, No.3, P.P 137-172, 1999.
K. Krishnoiger, S.H. Cheraghi, “Ant Algorithms: Review and Future
Application”.
S.J. Huang, “Enhancement of Hydroelectric Generation Scheduling
Using Ant Colony System Based Optimization Approaches”, IEEE
Trans. On Energy Conversion, Vol.16, No.3, Sep.2001.
D.Merkle, M. Middendorf and H. Schmech, “Ant Colony Optimization
for Resource Constrained Project Scheduling”, IEEE Trans. On
Evolutionary Computation, Vol.6, No.4, Aug.2002.
A.Bauer, B. Bullnheimer, R.F. Hartl and C. Strauss, “ A Ant Colony
Optimization Approach for Single Machine Total Tardiness Problem”,
IEEE 1999, P.P. 1445-1450.
N. S. Sisworahardjo and A.A. El Keib, “ Unit Commitment Using the
Ant Colony Search Algorithm”, IEEE 2002, P.P. 2-6.
T.Niknam, A.M. Ranjbar, and A.R. Shirani, “Optimization With Ant
Colony Algorithm”, BARGH Journal of Electrical Science and
Technology, No.36, P.P.11-19.
Y.H. Hou, Y.W. Wu, L.J. Li and X.Y. Xiong, “Generalized Ant Colony
Optimization for Economic Dispatch of Power Systems”, IEEE 2002,
P.P.225-229.
T.Niknam, H. Arabian and M. Mirjafari, “Reactive Power Pricing in
Deregulated Environments Using Novel Search Methods” is accepted
for presentation to IASTED International conference on PowerCon
2004, Greece.
T.Niknam and A.M. Ranjbar, “ Impact of Distributed Generation on
Distribution Load Flow”, International Power System Conference
Tehran, Iran, Oct. 2002.
Douglas Holtz-Eakin, “Prospects for Distributed Electricity Generation”,
September 2003.
“ Radial Distribution Test Feeders”, IEEE Trans. On Power Systems,
Vol.6, No.3, Aug.1991.
Bahman Bahmani was born in Shiraz, Iran. He received his B.S and M.S
degrees from Shiraz University and Sharif University of Technology
respectively. He is a member of faculty at the electrical engineering
department of Marvdaht Islamic Azad University. His interests include power
system modeling, power electronics, optimization methods, and evolutionary
algorithms.
Taher Niknam was born in Shiraz, Iran. He received his B.S, M.S and
PhD degrees from Shiraz University and Sharif University of Technology
respectively. He is a member of faculty at the electrical engineering
department of Shiraz University of Technology. His interests include power
system restructuring, impact of DGs on power system, power electronics,
optimization methods, and evolutionary algorithms.
Majid Nayeripour was born in 1971. He received his B. S. degree in
electronic Eng. from Guilan University and M.S degree in Electrical Eng.
from Esfahan University of Technology and PhD degree in Electrical Eng.
form Tarbiat Modares University, Tehran, Iran. Currently, he is an Assistant
Professor with the Shiraz University of Technology. His research interests
include FACTS devices, Power Quality and impact of DGs on power system.
113