Restore Power Losses using the Hybrid of the Minimum Spanning
Tree and Backward Forward Sweep
Meriem M’dioud
1a
, Rachid Bannari
1 b
and Ismail ELkafazi
2 c
1
Laboratory Engineering Sciences Ensa, Ibn Tofail University Kenitra, Morocco
2
Laboratory SMARTILAB, Moroccan School of Engineering Sciences, EMSI, Rabat, Morocco
Keywords: Backward Forward Sweep, Load Flow Analysis, Minimum Spanning Tree, Prim's Algorithm,
Reconfiguration, Radial Distribution Network, Power Losses, Voltage Profile.
Abstract: Reconfiguration of the electrical network is a famous tool still used to reduce losses. This method focused on
the changing of the topological state of the switching lines in the system. In this paper, the aim is to restore
the active power loss and upgrade the voltage profile at each node. This problem will be solved for the case
of the network without distribution generation units (DGs) and the case of the network with the presence of
the DGs because the injection of this last one at a non-optimal node gives rise to unnecessary losses and the
violation of the voltage out of the range limits. This reconfiguration will be done by using the prim’s
algorithm. Next, we apply the Backward Forward Sweep approach (BFS), aiming to check our proposed
constraints. By selecting the algorithm, total losses were chosen as an objective function by considering the
resistance of the edges as the weight of lines. The electrical network of 33 nodes with and without DGs was
presented to prove the efficiency of our proposed method. The simulation results prove that this algorithm is
perfect for finding good results (reduce losses, and improve voltage).
1 INTRODUCTION
For a few years, various electrical societies have been
guided towards the optimising of the unnecessary
expenses resulting from the higher losses in the
distribution system due to the lower voltage in this
section. This study based on the reconfiguration
strategy by changing the topological state of the
switching lines from 1- closed line to 0- open line and
vice versa. This study treated two cases: network with
and without the existence of the Distributed
Generation units (DGs). For this reason, in the next
part, we present much research concerning these two
kinds of networks.
For instance, in the case of the electrical network
without the presence of the DG units, we have the
study of the author of (Alvarez-Hérault & Marie-
Cécile, 2010), has select to use the problem of the
travelling salesman to find the new reconfiguration-
by using the Christofide algorithm. On another side,
the author of (Ahuja & Pahwa, 2005) have tried to
a
https://orcid.org/0000-0002-7855-3495
b
https://orcid.org/0000-0000-0000-0000
c
https://orcid.org/0000-0002-9921-0280
solve this problem by using the ant colony algorithm
due to its maturity to converge quickly and the
performance of the solution found. Furthermore, the
authors of (Enacheanu, 2008)have noted that the
genetic algorithm found the perfect results, by
optimizing the power loss, improving the voltage
profile, and minimizing the undistributed energy.
Besides, the authors of (Zhongfu Jiang, et al.,
2017) have studied the optimization of the annual
power generation cost and transmission cost; the
suggested algorithm is a mixed-integer linear
programming problem. Otherwise, it is important to
present the study of (Dodu, 1978) where the author
has chosen to solve the problem with the column
generation method to minimize the investment,
exploitation, and failure cost. Moreover, the authors
of (Leonardo W. Oliveira, et al., 2015) have chosen to
use the PSO algorithm which is an evolutionary
technique, aiming to restore the active power loss in
the radial electrical network. Even the authors of
(Hasmaini, Zalnidzham, Salim, Shahbudin, & Yasin,
108
M’dioud, M., Bannari, R. and Elkafazi, I.
Restore Power Losses using the Hybrid of the Minimum Spanning Tree and Backward Forward Sweep.
DOI: 10.5220/0010729400003101
In Proceedings of the 2nd International Conference on Big Data, Modelling and Machine Learning (BML 2021), pages 108-118
ISBN: 978-989-758-559-3
Copyright
c
2022 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
2019) have proposed to apply the graph theory by
using Kruskal’s algorithm to find the minimum
spanning tree, and they have compared their solution
with (Leonardo W. Oliveira, et al., 2015).
Always in
the same vision, we found, the study of the authors
(TD S. , 2017), where he considers the reactive power
as weights of edges to minimize losses, on the other
hand, the authors of (Tomoiaga, Chindris, Sumper,
Villafafila-Robles, & Sudria-Andreu, 2013) have
chosen the genetic algorithm that gives a perfect
quality at a lesser time.
As already presented, this study examines even
the network with the presence of the DG units. For
this reason, we introduce in the next part some studies
which solve this problem. However, it is important to
note that the DG is still a good technology that
provides power at or near the consumer, such as solar
panels and combined heat and power (Peng, 2004).
The integration of the DG units into the electrical
system helps to minimize losses on the transmission
and distribution lines (Carreno, Romero, & Padilha-
Feltrin, 2008). This technology is even applied to
reduce the quantity of power that must be produced at
the centralized power plants. Also, the DG units have
an important role in reducing the environmental
impacts resulting from the centralized generation
(Salama & El-Khattam, 2004).
To define the quantity of power demanded, the
author (Multon, 1999) has defined the forecasts of
power consumption. Furthermore, (Strasser, et al.,
2014) have analyzed the injection of renewable
energy sources, then they have followed a method to
control the energy and request the demand; at the end,
they have discussed the important role of the smart
grid to check the consistency between the consumer
demand and the supply of the electrical companies. In
the same vision, the author of (Caire, 2004) have
treated in their study the impacts of the injection of
the DG units on the quality of the electrical system
(risk of violation the voltage range limit), and it is
important to find a solution which allows the injection
of DG units in distribution systems with an elevated
injection rate. Otherwise, the authors of (Le Xie &
Marija D.Ilic, 2008) have proposed an optimal control
algorithm for distribution systems by applying the
Predictive Control Model to reduce the cost of
production.
To detect an optimal reconfiguration of the
network with the presence of the DG units, the
electrical companies have selected to search for new
strategies, aiming to enhance and minimize the
exploitation of the energy by reducing the losses.
In this side, we find the study of the author (Juma,
2018) who has selected the shark smell optimization
to minimize the total losses. Also, the authors
(Sivkumar Mishra, Debapriya Das, & Subrata Paul,
2014) have presented a simple algorithm for a
network with DG units by considering this last one as
a negative load. Moreover, the authors (Gallego,
Carreno, & Padilha-Feltrin, 2010) have discussed the
difference between the PQ bus and the PV bus, with
the first one is the case when the inserted power to the
network is considered as a negative load, but the
second one is where the reactive power of the DG
units depends on the voltage requested and the active
power injected is considered constant. Moreover, the
authors (Ogunjuyigbe, Ayodele, & Akinola, 2016)
have presented in their work the drawbacks resulting
of the insertion of the DG units on the features of the
network, such as the active power loss and the voltage
profile.
In the same issue, the authors (Chidanandappa,
Ananthapadmanabha, & H.C., 2015) have found a
new reconfiguration of the electrical system with the
presence of the DG units where the value of the losses
and the voltage profile are reduced by using the
combination of genetic algorithm and the backward/
forward sweep. Otherwise, the authors (Ahmad, Asar,
Sardar, & Noor, 2017) have examined in their study
the reliability analysis of the radial electrical system
in two cases (without and with the presence of the DG
units). Nevertheless, regarding the study of the
authors (Ma, Li, Zhang, Li, & han, 2017), where they
have selected to use the hybrid of the prim algorithm
and the particle swarm optimization algorithm aiming
to reduce losses in the presence of the DG units.
The authors (Hasmaini, Zalnidzham, Salim,
Shahbudin, & Yasin, 2019) and (TD s. , 2017) have
proposed that the weight of the line is the reactive
power to minimize losses. In this article, we have
used the resistance of edge as the weight of line due
to the effect of this factor on the power loss, and the
objectives are to reduce losses and improve voltage
profile.
To solve this issue, we have proposed to divide
the paper into five main sections. Section two
introduces the objective function and the constraints
of our problem. The third section gives the flowchart
of our proposed algorithm and presents case studies
used to perform our algorithm. Section four gives the
simulation results and presents a comparative study.
Furthermore, in the end, we conclude our study and
present the possible future research.
Restore Power Losses using the Hybrid of the Minimum Spanning Tree and Backward Forward Sweep
109
2 PROBLEMATIC
The origin of the problem comes at the peak demand,
is means where the electrical companies operate at
maximum; this last one gives rise to many losses that
give unnecessary expenses. Furthermore, the lower
value of voltage in the distribution network gives a
significant loss in this part of the network compared
with the transport network. When the load increases,
these losses increase. In this study, we examine the
issue of the reconfiguration of the distribution
network by using MATLAB software to find the
minimum spanning tree to automatically generate a
network structure with fewer losses.
2.1 Objective Function and Constraints
Our main objective is to minimize the active losses
cost, which is given by the equation:
Min K
𝑃
𝑗
= min K
mϵL
R
m
𝐼
m
2
(1)
Altenatively, K is the cost of the active losses;
equal to 100$/MWh (Hossein Moarrefi, 2013).
𝑃
𝑗
is
the total losses of the network, L is the set of network
lines, and
𝐼
m
is the current of the line m,
R
m
is the
current of the line m.
Where the k is a constant, so minimize cost is
mean to minimize the losses
mϵL
R
m
𝐼
m
2
Subject to the following constraints
a. Kirchhoff's law:
𝐼∗
𝐴
= 0
(2)
Where; I: row vector of current of each line of
network and A: incidence matrix of the network
b. Voltage range limit:
𝑉

𝑉
/𝑉

𝜀

(3)
𝑉

nominal voltage, 𝑉
is the voltage of bus j and
𝜀

is tolerance limit (Enacheanu, 2008) (+/-5% for
HTA and +6%/-10%BT)
c. current range limit :
𝐼
𝐼
,
(4)
𝐼
: current of edge m and 𝐼
,
: current
limit of line m.
To find the solution, we combine the prim’s
algorithm to find the minimum spanning tree and the
backward/ forward sweep to apply load flow aiming
to check the constraints (Saad Ouali, 2020).
2.2 Proposed Algorithm
In this paper, we consider the network as a graph, and
we use the graph theory to solve this problem by
using the prim’s algorithm; this last one is an
algorithm applied to search the minimum spanning
tree of a network. We select this method due to its
feature of high-speed switching circuits.
The prim’s algorithm helps to find the set of edges
that constitute a tree with minimum total weight also
connects all the nodes. In our case, it helps us to find
a tree with the total resistance of the tree is optimized.
To apply the prim’s algorithm, we start by
selecting a random initial node from the bus set. Then
we construct the tree by adding at each iteration the
line of the minimum weight that connects the tree to
another node. Figure 1 shows how the prim’s
algorithm operates.
Figure 1: Prim’s algorithm.
BML 2021 - INTERNATIONAL CONFERENCE ON BIG DATA, MODELLING AND MACHINE LEARNING (BML’21)
110
2.3 Backward Forward Sweep
This algorithm of the Backward Forward Sweep
(BFS) is used to check the constraints that the system
must respect. We choose this algorithm due to its
advantages: it easy to implement and uses simple
mathematic equations to find the current and the
voltage. In this paper, we have used the method of
BFS improved by the authors of (Saad Ouali, 2020).
To implement this method, the algorithm needs
two inputs: line data and load data. Also, this
algorithm is focused on three main steps:
1. First, we start from the last node.
2. After we used equation 5 to calculate the
nodal current: at each node “i” is calculated by:
𝐼
𝑐𝑜𝑛𝑗𝑃

𝑄

)/ 𝑉

),
i=1,2,…n
(5)
Where 𝑆
= 𝑃
+𝑗∗𝑄
is the power inserted at node
i, and 𝑉
is the voltage of bus i from iteration k.
3. Then we apply the Backward sweep: we
start from the last branch and do this calculation:
𝐽

 𝑐𝑜𝑛𝑗𝑃

𝑄

/
𝑉

) +
𝐽

, r=1,…
(6)
Where
𝐽

is current in edges downstream
node “i”.
4. Next, we apply the Forward sweep: we start
from the origin node and calculate the bus voltage by
employing the equation 7:
𝑉
𝑉

𝑍
𝐽
i=2,3,….,n
(7)
Where 𝑍
is the impedance of edges “i-1,i”
5. Go to step 2 and repeat all these steps until
code checks constraints described in the above
equations (2), (3), and (4).
2.4 Reduce Losses using Prim
Algorithm
Figure 2: Flowchart of the proposed algorithm.
Restore Power Losses using the Hybrid of the Minimum Spanning Tree and Backward Forward Sweep
111
Figure 2 presents our suggested method to
minimize losses. To apply this method, we need to
have the line and load data of the network (impedance
of edges, active power, and reactive power of the
buses) also the location and size of the DG units.
Our proposed algorithm is focused on three main
steps:
Step 1: start by initializing the line and load data.
Step 2: create the minimum spanning tree by using
the prim’s algorithm (already described – figure 1)
Step 3: if DG units exist, update the load data of
the bus where DG is connected and apply load flow
analysis described in section 2.3.
Else Run load flow analysis until the algorithm
checks the constraints.
2.4.1 Networks without DGs
To check the reliability of our proposed algorithm, we
apply the electrical distribution network of 33 buses.
The electrical and topological features of this network
are taken from the reference of (Baran & Wu, 1989),
the table 4 in appendices shows the line data and load
data of the standard electrical network.
Figure 3 presents the reconfiguration of the
network of 33 buses before reconfiguration. With the
total reactive power equal 2.3 Mvar, the losses value
before reconfiguration equal 15.17% from the total
power, and the total active power equal 3.715 MW.
The network works at the nominal voltage 12.66 kV,
and the apparent base power is 100 MVA.
Figure 3: Network IEEE ‘’ 33 bus before reconfiguration’’.
This electrical system is constituted from 33
buses, and 37 edges, where we have 32 close edges,
and 5 open edges. As shown in figure 4, the red lines
are the set of the open switches.
2.4.2 Network with DGs
The second case is the network with the existence of
the DG units, in this paper, we assume that we have 4
DGs. Table 1 gives the data of the DG units injected.
Table 1: DGs data (Hossein Moarrefi, 2013).
Location
(
bus
)
Size
(
MW
)
Power facto
r
28 0.1 0.95
17 0.2 0.95
2 0.14 0.98
32 0.25 0.85
As already described, the weight of the line, in this
case, is the resistance. So, the result before and after
the update of the line and load data is the same
because the injection of the DG has not any impact on
the resistance.
Otherwise, regarding the insertion of the DG
units, we must update the new active power and
reactive power of the node where the DG is injected,
then we apply the load flow analysis. For this reason,
we use the following equations 8, 9, and 10 (Seif,
2014).
P= P

- P

(8)
Q= Q

- Q

(9)
P

= a* Q

(10)
So, in this study, we consider the DG units as a
negative load, as shown in figure 4:
BML 2021 - INTERNATIONAL CONFERENCE ON BIG DATA, MODELLING AND MACHINE LEARNING (BML’21)
112
Figure 4: DG connected as a negative load to the bus.
After insertion of the DG unit, the new value of
the losses defined by equation 11:
𝑃

R
𝑃

P

+
𝑄

 𝑄

/𝑉
(11)
with:
R is the line resistance.
𝑃

presents the line losses.
P

is the active power consumption of the load.
Q

is the reactive power consumption power.
P

, Q

) present the active and reactive power
output of distributed generation.
a: is the power factor of DG.
To check the performance of our proposed
algorithm, a comparative study is done.
3 TEST AND RESULTS
3.1 Case without DGs
Figure 5: Test system in MATLAB.
Table 2: Comparative analysis.
Tie line Power
losses Mw
Minimum
p.u.Voltage
Bus No of
minimum p.u.
volta
g
e
Time S
Base case 33- 34-35-36-37 0.2024 0.7717 33 -
Prim’s (proposed) 12-27-33-34-35 0.12489 0.95133 25 0.2235
Prim’s (TD s. , 2017) 7-8-13-29-37 0.1165 0.9446 30 -
Kruskal’s (Mohamad, 2019) 16-27-33-34-35 0.1786 0.9282 17 0.8566
BPSO & MSSO (JUMA S. A., 2018) 7-9-14-32-37 0.139 0.9479 18 34.63
Redefined Genetic 7-10-14-36-37 0.2007 0.8330 33 -
Restore Power Losses using the Hybrid of the Minimum Spanning Tree and Backward Forward Sweep
113
Figure 6: profile tension improvement.
After implementing of our line and load data, the new
reconfiguration is 12-27-33-34-35, as shown in figure
5.
Table 2 presents the comparative study, and it is
noticed that when we use our proposed algorithm, the
losses are reduced and becomes 0.1248 Mw, instead
of 0.2024 MW for the standard case and 0.178 MW
for the case of new architecture using the Kruskal
algorithm, 0.139 MW for the BPSO algorithm, and
0.2007 MW for the case of reconfiguration by using
the redefined genetic algorithm.
Furthermore, table 2 indicates that the minimum
voltage of our proposed algorithm is 0.95133 p.u. at
node 18, this value respects the range limit of voltage.
Also, it is improved compared with the standard case
where the minimum voltage equal 0.7717 p.u. for the
Kruskal’s algorithm, the minimum voltage is 0.9282
p.u., for the BPSO it is noticed that the minimum
voltage is equal 0.9479p.u., and for the redefined
genetic, the minimum voltage is 0.8330p.u. So, we
conclude that our suggested method improves the
voltage profile better than the other recent studies.
Figure 6 and Table 2 present the performance of
the suggested prim’s algorithm compared with other
recent articles. The table in appendix B gives the
value found in each bus before and after the
restoration of active power loss by using the
reconfiguration of the network without DGs.
Finally, we conclude that our proposed algorithm
helps to minimize the active power loss and enhance
the voltage profile better than the other studies. Also,
this algorithm gives result in this case after 0.2235s,
this value it is lesser than the Kruskal's algorithm that
finds result after 0.8566s and lesser than the BPSO
algorithm that takes 34.64s. Appendix B gives the
voltage profile of each bus for the standard case and
our result.
3.2 Case with DGs
As presented previously, we consider that the weight
of edges is their resistance. Furthermore, the insertion
of the DG has not any impact on the impedance. So,
the implementation of the line data, in this case, gives
the same reconfiguration of the case of the network
without the presence of the DG after applying the
prim’s algorithm, as shown in figure 5. After this step,
we update the load data of the bus where the DG is
injected by using the equation (8), (9), and (10). Then
we apply the load flow method by using the main step
of the backward/ forward sweep (equation (5), (6),
and (7)).
Table 3 presents the simulation results of our
proposed algorithm in this case and the results of
other recent studies. Consequently, it is noted that the
losses found using our proposed algorithm equal
0.1331 MW; this value is big than the losses found by
(Hossein Moarrefi, 2013), where they have found
0.1241MW. However, it is important to note that the
losses of the standard case equal 0.2465 MW, and this
value is very big than our case.
Considering the voltage profile, our proposed
method enhances the voltage at each node better than
the base case and the (Hossein Moarrefi, 2013). Also,
the minimum voltage of our case equals 0.95025 p.u.
Nevertheless, in the case of the authors (Hossein
Moarrefi, 2013), the minimum voltage is 0.9124 p.u.
and 0.8950 p.u. for the base case. Furthermore, our
proposed algorithm gives the results after 0.2235 s.
Figure 7 presents the voltage profile curve in two
cases: the standard case and the case of our proposed
algorithm. It is noticed that the value of the voltage of
each node using our proposed algorithm is more
improved than the base case.
BML 2021 - INTERNATIONAL CONFERENCE ON BIG DATA, MODELLING AND MACHINE LEARNING (BML’21)
114
Table 3: Comparative analysis for case network with DG
Base case with
DG
GA with DG (Hossein
Moarrefi, 2013)
Proposed prim’s
algorithm with DG
Open branches 33-34-35-36-37 12-15-18-21-22 12-27-33-34-35
The node of minimum voltage 18 - 25
Minimum voltage profile (p.u) 0.8950 0.9124 0.95025
Total losses (MW) 0.2465 0.1241 0.1331
Figure 7: voltage profile for radial system distribution with DGs.
So, considering the improvement of the voltage
profile, our study is perfect than the base case and the
study (Hossein Moarrefi, 2013). Table 5 presents the
value of the voltage before and after reconfiguration
using our method.
4 CONCLUSION
In this article, our objective is to restore the losses
aiming that the electrical plants can follow the
customer's request. So, for this reason, this study
proposes to use a combination of the prim’s algorithm
and the backward/ forward sweep due to their
advantages to find the minimum spanning tree and to
check the constraint with a higher speed to converge.
For the Prim’s algorithm, we have considered that the
weight of edges is the resistance of lines also is used
to check the constraint of the network have a radial
topology, for the backward/ forward sweep is used to
check the constraints of the current and voltage range
limit also the first law of Kirchhoff.
To check the quality and the reliability of our
study, we have used the electrical distribution
network of 33 buses for two cases: network without
and with the presence of the DG units. The simulation
results of these two cases show that our hybrid
method is perfect for minimizing losses and
enhancing the voltage at a lesser time compared with
other studies. For the next work, we advise using an
evolutionary algorithm to solve the problem aiming
to find the global optimum.
REFERENCES
Ahmad, S., Asar, A. U., Sardar, S., & Noor, B., 2017.
Impact of distributed generation on the reliability of
local distribution system. IJACSA International
Journal of Advanced Computer Science and
Applications, 8(6), 375-382.
Ahuja, A., & Pahwa, A., 2005. Using Aint Colony
Optimization for loss minimization in distribution
networks. Proceedings of the 37th Annual North
American Power Symposium, IEEE.
Alvarez-Hérault, & Marie-Cécile., 2010. Architectures des
réseaux de distribution du futur en présence de
production décentralisée, Sciences de l'ingénieur.
Grenoble: Institut National Pplythechnique de
Grenoble.
Baran, M. E., & Wu, F. F., 1989. Network reconfiguration
in distribution systems for loss reduction and load
balancing. IEEE Transactions on Power Delivery, 4(2),
1401 - 1407.
Restore Power Losses using the Hybrid of the Minimum Spanning Tree and Backward Forward Sweep
115
Caire, R., 2004. Gestion de la production décentralisée
dans les réseaux de distribution. tel-00007677: Institut
National Polytechnique de Grenoble.
Carreno, E. M., Romero, R., & Padilha-Feltrin, A., 2008.
An efcient codifcation to solve distribution network
reconfguration for loss reduction problem. IEEE
Transactions on Power Systems, 23(4), 1542-1551.
Chidanandappa, R., Ananthapadmanabha, T., & H.C., R.,
2015. Genetic algorithm based network reconfiguration
in distribution systems with Multiple DGs for time
varying loads. SMART GRID Technologies, 21, 460-
467.
Dodu, J., 1978. Modèle dynamique d’optimisation à long
terme d’un réseau de transport.
http://www.numdam.org/item?id=RO_1978__12_2_1
41_0.
Enacheanu, F. B., 2008. outils d’aide à la conduite pour les
opérateurs des réseaux. HAL https://tel.archives-
ouvertes.fr/tel-00245652.
Gallego, L. A., Carreno, E., & Padilha-Feltrin, A., 2010.
Distributed generation modeling for unbalanced three-
phase power flow calculations in smart grids.
Transmission and Distribution Conference and
Exposition: Latin America (T&D-LA).
Hasmaini, M., Zalnidzham, W. I., Salim, N. A., Shahbudin,
S., & Yasin, Z. M., 2019. Power system restoration in
distribution network using kruskal algorithm.
Indonesian Journal of Electrical Engineering and
Computer Science, 16(1), 8.
Hossein Moarrefi, M. N., 2013. Reconfiguration and
distributed Generation(DG) placement considering
critical system condition . 22nd International
Conference on Electricity Distribution . stockholm.
JUMA, S. A., 2018. Optimal radial distribution network
reconfiguration using modified shark smell
optimization. http://hdl.handle.net/123456789/4854.
Le Xie, & Marija D.Ilic., 2008. Model Predictive Dispatch
in Electric Energy Systems with Intermittent
Resources. IEEE International Conference on Systems,
Man and Cybernetics. IEEE.
Leonardo W. Oliveira, Edimar J. Oliveira, Ivo C. Silva,
Flávio V. Gomes, Thiago T. Borges, André L. M.
Marcato, & Ângelo R. Oliveira., 2015. Optimal
restoration of power distribution system through
particle swarm optimization. IEEE Eindhoven
PowerTech.
Ma, C., Li, C., Zhang, X., Li, G., & han, Y., 2017.
Reconfiguration of Distribution Networks with
Distributed Generation Using a Dual Hybrid Particle
Swarm Optimization Algorithm. Hindawi
Mathematical Problems in Engineering, 2017, 11.
Mohamad, H., 2019. Power system restoration in
distribution network using. Indonesian Journal of
Electrical Engineering and Computer Science, 16(1), 8.
Multon, B., 1999. L’énergie électrique: analyse des
resources et de la production. Journées de la section
électrotechnique du club EEA.
Ogunjuyigbe, A., Ayodele, T., & Akinola, O., 2016. Impact
of distributed generators on the power loss and voltage
profile of sub-transmission network. Journal of
Electrical Systems and Information Technology, 3, 94-
107.
Peng, F. Z., 2004. Editorial Special Issue on Distributed
Power Generation. IEEE TRANSACTIONS ON
POWER ELECTRONICS, 19(5), 2.
Saad Ouali, A. C., 2020. An Improved Backward/Forward
Sweep Power Flow Method Based on a New Network
Information Organization for Radial Distribution
Systems. Hindawi Journal of Electrical and Computer
Engineering, 2020(ID 5643410), 12.
Salama, M. M., & El-Khattam, W., 2004. Distributed
generation technologies,defnitions and benefts. Electric
Power Systems Research, 71(2), 119-128.
Seif, M. A., 2014. Optimal voltage control and loss
reduction in microgrid by active and reactive power
generation. Journal of Intelligent & Fuzzy Systems, 27,
1649–1658.
Sivkumar Mishra, Debapriya Das, & Subrata Paul., 2014.
A Simple Algorithm for Distribution System Load
Flow with Distributed Generation. IEEE International
Conference on Recent Advances and Innovations in
Engineering. Jaipur, India: IEEE International
Conference on Recent Advances and Innovations in
Engineering.
Strasser, T., Andrén, F., Kathan, J., Cecati, C., Buccella, C.,
Siano, P., Mařík, V., 2014. A Review of Architectures
and Concepts for Intelligence in Future Electric Energy
Systems,. IEEE Transactions on Industrial Electronics,
62(4), 2424 - 2438.
TD, s., 2017. loss minimization in distribution network by
using prim’s algorithm. International Journal of
Applied Engineering Research , 9(24), 15.
Tomoiaga, B., Chindris, M., Sumper, A., Villafafila-
Robles, R., & Sudria-Andreu, A., 2013. Distribution
system reconfiguration using genetic algorithm based
on connected graphs. Electric Power Systems Research
journal, 104(216-225), 10.
Zhongfu Jiang, Donglei Sun, Long Zhao, Xiaoming Li,
Shan Li, & Xueshan Han., 2017. A new method for
reference network considering network topology
optimization. 2017 Chinese Automation Congress
(CAC), IEEE.
BML 2021 - INTERNATIONAL CONFERENCE ON BIG DATA, MODELLING AND MACHINE LEARNING (BML’21)
116
APPENDIX
Table 4: line data and load data of IEEE 33 bus (Baran & Wu, 1989).
Line data Load data
Branch N° From bus To bus R (ohm) X(ohm) Pl(Kw) Ql(kvar)
1 1 2 0,0922 0,047 100 60
2 2 3 0,493 0,2511 90 40
3 3 4 0,366 0,1864 120 80
4 4 5 0,3811 0,1941 60 30
5 5 6 0,819 0,707 60 20
6 6 7 0,1872 0,6188 200 100
7 7 8 0,7114 0,2351 200 100
8 8 9 1,03 0,74 60 20
9 9 10 1,04 0,74 60 20
10 10 11 0,1966 0,065 45 30
11 11 12 0,3744 0,1238 60 35
12 12 13 1,468 1,155 60 35
13 13 14 0,5416 0,7129 120 80
14 14 15 0,591 0,526 60 10
15 15 16 0,7463 0,545 60 20
16 16 17 1,289 1,721 60 20
17 17 18 0,732 0,574 90 40
18 2 19 0,164 0,1565 90 40
19 19 20 1,5042 1,3554 90 40
20 20 21 0,4095 0,4784 90 40
21 21 22 0,7089 0,9373 90 40
22 3 23 0,4512 0,3083 90 50
23 23 24 0,898 0,7091 420 200
24 24 25 0,896 0,7011 420 200
25 6 26 0,203 0,1034 60 25
26 26 27 0,2842 0,1447 60 25
27 27 28 1,059 0,9337 60 20
28 28 29 0,8042 0,7006 120 70
29 29 30 0,5075 0,2585 200 600
30 30 31 0,9744 0,963 150 70
31 31 32 0,3105 0,3619 210 100
32 32 33 0,341 0,5302 60 40
Tie Lines
33 8 21 2 2
34 9 15 2 2
35 12 22 2 2
36 18 33 0,5 0,5
37 25 29 0,5 0,5
Restore Power Losses using the Hybrid of the Minimum Spanning Tree and Backward Forward Sweep
117
Table 5: Bus voltage before and after reconfiguration without DGs.
Network without DGs Network with DGs
Bus NO Voltage profile before
reconfiguration
Voltage profile after
reconfiguration
Voltage profile before
reconfiguration
Voltage profile after
reconfiguration
1 1 1 1 1
2 0.997 0.9975 0.9969 0.9975
3 0.9829 0.9857 0.9821 0.9856
4 0.9754 0.9828 0.9741 0.9827
5 0.968 0.9802 0.9662 0.9701
6 0.9496 0.9743 0.9458 0.9742
7 0.9461 0.9723 0.9413 0.9722
8 0.9324 0.9700 0.9357 0.9699
9 0.9261 0.9680 0.9279 0.9679
10 0.9203 0.9665 0.9205 0.9664
11 0.9194 0.9664 0.9194 0.9663
12 0.9176 0.9662 0.9176 0.9661
13 0.9131 0.9659 0.9091 0.9658
14 0.911 0.9651 0.9056 0.9650
15 0.9095 0.9648 0.9031 0.9647
16 0.908 0.9648 0.9007 0.9644
17 0.9058 0.9638 0.8957 0.9610
18 0.9052 0.9632 0.8951 0.9604
19 0.9965 0.9969 0.9964 0.9969
20 0.9929 0.9931 0.9928 0.9934
21 0.9922 0.9927 0..9921 0.9927
22 0.9916 0.9920 0.9915 0.9920
23 0.9793 0.9782 0.9785 0.9779
24 0.9727 0.9630 0.9718 0.9623
25 0.9693 0.9513 0.9685 0.9503
26 0.9477 0.9741 0.9438 0.9740
27 0.9451 0.9739 0.9412 0.9738
28 0.9337 0.9735 0.9290 0.9733
29 0.9255 0.9696 0.9204 0.9693
30 0.922 0.9606 0.9168 0.9597
31 0.9178 0.9596 0.9120 0.9581
32 0.9169 0.9582 0.9108 0.9566
33 0.9166 0.9579 0.9105 0.9563
BML 2021 - INTERNATIONAL CONFERENCE ON BIG DATA, MODELLING AND MACHINE LEARNING (BML’21)
118