Beruflich Dokumente
Kultur Dokumente
NOMENCLATURE
FL
b branch resistance.
b branch complex current.
Current vector of branches and maximum current of
branches.
Vector of node currents.
Incidence matrix.
Node voltage.
Node minimum voltage.
Node maximum voltage.
Radial net branch number.
Node number.
Source number.
Fundamental loop.
Total branch number.
Manuscript received May 25, 2005; revised November 3, 2005. This work
was supported by the National Commission for Investigation, Science and Technology of Chile (CONICYT). Paper no. TPWRS-00318-2005.
J. Mendoza and E. Lpez are with the Department of Electrical Engineering,
University of Concepcin, Concepcin, Chile (e-mail: jorgemendoza@udec.cl;
elopez@udec.cl).
R. Lpez is with the Faculty of Law and Economics, University Paris XI,
Paris, France (e-mail: rlopezg@ieee.org).
D. Morales and P. Dessante are with the Dpartement Electrotechnique et
Systmes dEnergie, Suplec, Paris, France (e-mail: dario.morales@supelec.fr;
philippe.dessante@supelec.fr).
R. Moraga is with the Engineering and Odontology School, Desarrollo University, Concepcin, Chile (e-mail: rmoraga@udd.cl).
Digital Object Identifier 10.1109/TPWRS.2006.873124
I. INTRODUCTION
and mutation operators. This allows converging to the best solution [14][16].
The first work that applied GA to the reconfiguration problem
was developed in 1992 by Nara [17]. In spite of the excellent
results, the conclusion of this paper and the study developed
by Sarfi [18] pointed out the need for computers with greater
processing speed. The principal trouble presented in [17] is
related to the binary codification used; it identified the arc
(branch) number that contains the th open switch and identifies
the switch that is normally open in this arc. That codification
type can be very long, and it grows in proportion with the
switch number. Also, an approximated fitness function was
used to represent the system power loss.
Nevertheless, the technological advances in computer hardware allow today the application of these methodologies with
greater benefits and fewer limitations in terms of the problem
dimensions and computational times involved in the population
size, number of generations, and objective function evaluation.
In [23], the GA method is refined, in the reconfiguration
problem, by modifying the string structure and fitness function.
Here, a binary string represents only positions of the open
switches in the distribution network. Consequently, the length
of the string is reduced, depending on the number of open
switches. The fitness function also considers constraints of the
systems. An adaptive mutation process is used to change the
mutation probability.
Lopez in [24] made an important contribution to the reconfiguration process using GA, with excellent results in the
simulation times, when introducing graph compression, current
flows analysis, stochastic minimum extension trees, and diakoptic compensatory currents. All of these techniques are used
to simplify the systems model, allowing the evaluation of the
objective function with less computational efforts. Also, a set
of filters is introduced in order to eliminate the individuals that
transgress the system operational constraints (like over currents
and voltage ranges). However, the computational efforts are
still significant due to the generated random population, where
many do not override the radiality and connectivity filters.
In [26], a GA technique is applied to the multiobjective reconfiguration problem. This work used the Prufer number
codification, to avoid a tedious meshed check algorithm. The
crossover is applied in two points randomly selected. The single
point mutation is applied, and the roulette wheel approach is
used to select the rest of the individuals.
The work presented in [27] is the only one that develops a
method to create a feasible population. This method is based on
the concept called path-to-node. This scheme is based on the
preliminary identification of alternative paths linking each bus
to the substation. This path definition creates radial topologies
for the initial population with excellent results but only using
traditional genetic operators.
As a whole, the great majority of the GA applications to the
reconfiguration problem are done using a binary codification
that represents the location and codifications of the switches in
the system. In real systems, the above may result in string that is
too long, decreasing the efficiency in the search for the optimum.
An attempt to solve this problem has been done by developing
new codification strategies (keeping on with a binary line) and
949
by new ways of improving the convergence through the probability of adaptive mutation
For this reason, in this paper, a GA addressed population generation criteria that avoids nonfeasible individuals from a network structure standpoint is used. This changes drastically the
way of using GA in the reconfiguration process. This is based
on the work of Lin in [20], where the switching indexes of the
network were analyzed, to determine the opening of the control
devices.
II. SOLUTION METHOD
A. Minimal System Power Loss Problem
Reconfiguration is encompassed in problems of planning and
operation of primary distribution systems. The principal objective is to find a radial operating structure that minimizes the
system power loss while satisfying operating constraints. The
mathematical formulation for the minimization power loss reconfiguration problems is presented in the literature in different
ways. In this paper, the problems formulation is presented as in
[12]
minimize
(1)
subject to
(2)
(3)
(4)
(5)
Equation (1) corresponds to the objective function to minimize and represent the total power loss of the system. Equation (2) corresponds to the matrix of nodal load current balances. Equation (3) corresponds to feeder thermal limits and to
the maximum capacity of substations. Equation (4) considers
voltage constraints in each node. Equation (5) describes radiality constraints of the primary distribution system.
Up until now, the approach to the reconfiguration problem
using GA relied on the use of filter and/or mechanisms to create
initial populations in order to avoid the evaluation of nonfeasible
individuals. This paper considers the application of system fundamental loops to make decisions about how to create feasible
individuals and overall how to obtain feasible individuals later
to apply the genetic operators.
B. Reconfiguration Using Fundamental Loops
The proposed methodology creates feasible topologies using
topological analysis. It is made to identify the fundamental
closed loops of the system in order to originate radial topologies.
When analyzing meshed networks, the number of fundamental loops (FLs) is
FL
(6)
950
Loop
Loop
Loop
In order to create a radial topology, one should select from
the group of fundamental loop vector elements to be disconnected (one for loop). It is important that the vectors elements
are not repeated (or have the same genetic information) in a selected topology. The combination of elements can be generated
with these vectors, creating all possible radial topologies of the
system.
For the system shown in Fig. 1, the total number of feasible
radial topologies, using the proposed method, is 30, while 56
topologies were observed using a random methodology. This is
a great advantage of the proposed method. These considerations
allow the proposed GA to limit the generation of nonfeasible individuals. This also reduces the combinatorial searching space.
C. Genetic Algorithms
GA is a technique based on the Theory of Evolution. It can
be applied to a wide range of engineering problems.
The genetic function principally works at the genotype level,
but they can also function at the phenotype level. The principal
operator is the crossover, and the secondary operator is mutation. The way to select individuals is probabilistically based on
the individual fitness.
The GA methodology structure is as follows:
1) codification of individuals (topologies);
2) generation of a feasible initial population;
3) each individual (solution) evaluated through the fitness
function (system power loss);
4) application of genetic operators;
5) repetition of the third and fourth steps, until reaching the
total number of generations.
The binary codifications for [17], [22], [24], and [25] grows in
proportion with the switch number of the system (furthermore,
when the systems are larger and complex). According to the
codification presented in [23], it clearly reduces the number of
bits used. Nevertheless, if the amount of switch increases, a big
number of bits for the codification may be needed.
In this paper, each position (gene) of the string represents a
randomly selected element of each fundamental loops vector.
proposed method
The main advantage of this codification is based on creating
individuals guided through the fundamental loops vector, allowing to produce radial topologies that, with the nonguided binary codifications, are difficult to achieve
Thus, the individual selection process is done through a probabilistic tournament of uniform distribution [22].
2) Application of Genetic Operators: The great majority of
the GA applications to the reconfiguration problem use the traditional mutation and crossover techniques [17], [21], [22], [24],
[26]. Only in some cases, efforts have been done to do the following:
choose out of traditional crosses the one more suitable for
the reconfiguration problem [19];
develop a process to change the mutation probability [23],
[25].
951
Fig. 5.
Fig. 6.
952
TABLE I
SYSTEM PARAMETERS
TABLE II
RESULTS OF RECONFIGURATION FOR THE PROPOSED METHOD
TABLE III
RESULTS OF RECONFIGURATION FOR OTHER REFERENCES
953
TABLE IV
RESULTS OF [24]
TABLE V
RESULTS OF THE PROPOSED METHODOLOGY
Fig. 10.
Fig. 11.
In this paper, a new methodology for minimal loss reconfiguration using GA is presented. This technique is based on
the construction of an initial population of feasible individuals
using the system loops and applying specialized genetic operators of accentuated cross and directed mutation. This reduces
the searching space, analyzing only feasible radial topologies.
The proposed method changes drastically the way of using
GA in the reconfiguration process, allowing the optimal reconfiguration of large distribution systems, with less computational
efforts (minimizing the required memory and CPU time), using
a simple code, and overall improvement of the searching ability.
When a small number of switching elements is considered,
the proposed methodology uses a smaller searching space (eliminating from the fundamental loops vector those lines that do
not have switching elements), simplifying the problem resolution and reducing (even more) the CPU time involved.
Consequently, this paper proposes a formal and robust
strategy for approaching the large minimal loss reconfiguration problems using GA. Better solutions than those attained
through other techniques, such as GA or traditional algorithms,
are found.
REFERENCES
[1] A. Merlin and G. Back, Search for minimum-loss operational spanning
tree configuration for an urban power distribution system, in Proc. Fifth
Power System Conf. (PSCC), Cambridge, U.K., 1975, pp. 118.
[2] D. Shirmohammadi and H. W. Hong, Reconfiguration of electric distribution networks for resistive line losses reduction, IEEE Trans. Power
Del., vol. 4, no. 2, pp. 14921498, Apr. 1989.
[3] S. Civanlar, J. Grainger, H. Yin, and S. Lee, Distribution feeder reconfiguration for loss reduction, IEEE Trans. Power Del., vol. 3, no. 3, pp.
12171223, Jul. 1988.
[4] M. E. Baran and F. Wu, Network reconfiguration in distribution system
for loss reduction and load balancing, IEEE Trans. Power Del., vol. 4,
no. 2, pp. 14011407, Apr. 1989.
[5] S. Goswami and S. Basu, A new for the reconfiguration of distribution
feeders for loss minimization, IEEE Trans. Power Del., vol. 7, no. 3,
pp. 14841491, Jul. 1992.
954
Enrique Lpez was born in Lota, Chile. He received the Electrical Engineer degree in 1974 from
Universidad Tcnica del Estado, Estado, Chile, and
the Ph.D. degree in 1983 from Institut National Politechnique de Grenoble (INPG), Grenoble, France.
Currently, he is an Associate Professor in the
Electrical Engineering Department, Universidad
de Concepcin. His interest areas are planning,
optimization, control, reliability, and quality of
electrical systems.