You are on page 1of 4

2012 3rd International Conference on System Science, Engineering Design and Manufacturing Informatization

A Study of Distribution Network Fault Location including Distributed Generator

Based on Improved Genetic Algorithm

Youjun Yue Yunlei Zhao


Tianjin University of Technology Tianjin University of Technology
Tianjin Key Laboratory for Control Theory & Applications Tianjin Key Laboratory for Control Theory & Applications
in Complicated Systems in Complicated Systems
Tianjin, China Tianjin, China
bakeryueyj@163.com zhaoyunlei0312@sina.com
Hui Zhao Hongjun Wang
Tianjin University of Technology Tianjin University of Technology
Tianjin Key Laboratory for Control Theory & Applications Tianjin Key Laboratory for Control Theory & Applications
in Complicated Systems in Complicated Systems
Tianjin, China Tianjin, China
zhaohui3379@126.com hongewang@126.com

Abstract: In the distribution network fault section positioning, will be uploaded from FTU to SCADA. According to this
the standard genetic algorithm can be used effectively. But the information, fault can be confirmed.
network including Distributed Generator power supply has its There are two major algorithm of fauit location for
own characteristics, and the genetic algorithm appear some distribution network:direct and indirect.Typical direct
defects. Therefore, aiming at accurate positioning of faults in algorithm is the matrix algorithm which fixes fault position
the network including DGs, the genetic algorithm coding and in the basis of fault judgement matrix that formed by
fitness function are improved in the paper. Through network topology and fault information matrix.It has a
simulation, the effectiveness of the proposed algorithm is strong point on calculation rate,but the fault tolerance is
verified.
poor.When the information from FTU is error,the fault
Keywords:distributed generator, genetic algorithm,fault position also maybe fixed error.In practice,FTUs are
location. outdoors,therefore,the information loss and distorition often
occurs.Then the fault position can’t be fixed exactly and
security and reliable of the distribution network can’t be
I. INTRODUCTION guarantee.Indirect algorithm is consist of various heuristic
search algorithm,as genetic algorithm and ant colony
With the introduction of the distributed generator (DG), algorithm.The latter one is through constricting to optimal
traditional distribution network gradually developed into a path using the accumulation and update of information
more complicated power network and the direction of the element,but the algorithm will be very slow because of
trend is random changed.When put the DG to feeder, the information element’s lack on initial stage.Using genetic
traditional distribution network protection which is based on algorithm on multiple power supply system,positive
radiation network current protection can’t adapt to the new direction of the network is required setting repeatedly,so the
operating environment. IEEEStd.1547 rules that when efficiency of fault location is enormously reduced.For this
distribution network’s fault occurs, DG must be cut off first reason,the generator algorithm is required to be improved in
to ensure correct operation of the relay protection. This rule order to realize fault position fixed reasonably and exactly
restricted the DG's normal operation for some degree,and under the condition of DGs’ accessing.[2][3]
also weaken the advantage of distributed generator on
improvement of the power supply reliability.Therefore,based II. SWITCH FAULT CURRENT’S NEW DEFINITION AND
on the present current protection and automation CODING PROBLEM
communication system, don't cut off DG and make fault
located fast and effectively have a great role to the
promotion of distributed power generation.In the When Genetic algorithm is used in the traditional
engineering practice,feeders generally have no voltage distribution network fault location, if switch to FTU
transformers,so researching a fault location algorithm which monitoring point detects a over-current and its direction is
is based on electrical flow has the practical significance.[1] the same as defined in the positive direction, in fitness
The distribution networks are equipped with feeder function the fault current status of switch Ij value for
terminal units (FTU) and supervisory control and data ‘1’.Other situations Ij is ‘0’.This definition does not apply
acquisitions (SCADA).When fault occurs, the current status with multiple power distribution network, so the text will
redifine Ij.
978-1-4673-0915-8/12/$31.00 ©2012 IEEE

103
For a switch, confirm its upstream power supply first downstream line for the line between the switch and its
which is from the switch recently.Other power sources are downstream.Different from single power supply
its downstream power supply.This paper defines the positive system,when more than one power sources supply for a
direction of a switch is from upstream power to downstream switch,the current through this switch has relationship with
power. every power source that supply power.Therefore,the paper
When FTU detects that the direction of fault defined a new switch function,as shown in (1).
over-current is accordant with the assumed positive direction N M1 M2

of this switch, the switch status value Ij is ‘1’.If these two I *j (s) (–KDGi ) u[0  (–s ju )]  (–s jd ) (1)
directions is opposite,Ij will be ‘-1’.When there is no fault i 1 u 1 d 1

over-current,Ij is ‘0’.Using code ‘-1’ for fault current status *


In (1), I ( s ) is the switch function of j which is
j
of the switch is because of that the power system include of
DG is different from single power radialized feeder structure. associated with the status of line and it is a expected status.
The fault status of line between two switches also codes KDGi represents whether DG supply power for system. When
‘1’,and the normal status of the line codes ‘0’. the DG is operating, its status code ‘1’.If not, code ‘0’. sju is
As shown in Fig.1 distribution network structure the status of the upstream line of switch j and sjd is the status
diagram,the upstream power source of switch S1,S2 and S3 of the downstream line of switch j. When the line is
can select the main power supply S,and the downstream will breakdown, the value is ‘1’ and if the line run normally, the
M1
be DG1,DG2.Similarly,the upstream power source of switch
S4,S5 and S6 is DG1,and the downstream power sources are value is ‘0’. –s
u 1
ju is the logic or operation of the fault
S and DG2.Switch S7,S8 and S9’s upstream power source is
M2
DG2,and the downstream power sources are S and DG1.S4
as an example,its positive direction is from upstream power status of switch j on the upstream line. Similarly, –s
d 1
jd is
source DG1 to downstream power sources S and DG2.When
fault K1 take place on line c,the direction of fault the logic or operation of the fault status of switch j on the
N
over-current through S4 is accordant with assumed positive
direction,so the status of switch is ‘1’.When the fault takes downstream line. –K DGi is ‘0’ when there is no DG
place on line d,the over-current of S4 are supplied by S and i 1
DG2 and its direction is opposite with the assumed positive *
operating in distribution network and I j ( s ) is same with
direction.So the status of S4 is ‘-1’.When there is no fault in
the network,the status of switch is ‘0’.The status of other the switch function of single power source system.So the
switches is determined similarly with S4.When fault K1 new switch function has adaptability for changing Topology
occurs,the status of line c codes ‘1’ and the status of other network structure.
nomal lines is ‘0’.
IV. THE STRUCTURE OF THE FITNESS FUNCTION

Use genetic algorithms on network fault location to the


distribution networkˈis finding fault line in the solution
space which include all the possible fault lines. According to
this idea, and the switch function which has constructed
already above, Fitness function F(s) can be structed. As
shown in (2).
k
F (s) ¦
j 1
I j  I *j ( s ) (2)

Figure1 Containing DG simplified distribution network In formula (2), k is the total number of the switches; s
is the state vector of the feed line. It consists of states of all
feed lines. When a feed line is on fault state, it is indicated
III. ESTABLISH INPROVED SWITCH FUNTCTION by element ‘1’. When it is in normal status, it is indicated
by element ‘0’.Ij is fault current limit signal of the switch j,
If the fault line is fixed by the switches’ information of they are actual state of the switches which feedback to the
fault over-current that analysed by genetic algorithm,a SCADA control center by FTU. As to fitness function, it is
transition which is from the status of fault line to the a known quantity. When the direction of current
over-current status of switches must be constructed.This information is the same as the positive direction of the
transition can be realized by a function,and it is switch switch, it is indicated by element ‘1’, and it is indicated by
function that reflects a incidence relation between lines and element ‘-1’ on the opposite. Then it is indicated by
*
switches.[2] element ‘0’ when it has none current information. I j ( s )
For a switch, define the upstream line which is between
is switch function which can be calculated by formula (1).
this switch and its upstream source. Similarly, define the
The desired status which is deduced by the actual

104
feeders state and switch function is not always accordant
with the actual switch fault over-current signal status of FTU.
Therefore, the process of solving the fitness function is to
search for the optimal solution in the solution spacethe
process, which is the process to find the smallest solvable
group of equation (2). The value (Numeric string or
chromosome) of this solution corresponds with the feeder
segment that is the actual feeder segments.
In fact, the fitness function of formula (2) is defective,
and paragraphs on the line which from fault point to power
point may be mistaken. The text adds an additional entry in
the right side of the formula (2), the improved fitness
function as shown in formula (3).
k M
F (s) ¦
j 1
I j  I *j ( s )  ¦s
i 1
i Z (3)

According to the famous concept of minimum set of the


theory of fault diagnosis, that is: Select the solution has
minimum number of fault lines in the fault diagnosis results.
So plus on original fitness function to prevent mistaking. In
which it’s the summation of all fault states. At the same time,
in this paper, to prevent mistakes, plus Z (0< Z <1) to
prevent mistakes.[4] Figure 2 Flow diagram

V. THE REALIZATION OF THE GENETIC ALGORITHM VI. AN EXAMPLE SIMULATION ANALYSIS

All switch's FTU will be uploading switch's fault Figure 1 as an example, when fault K3 occurs, there are
current bounds signal immidiately when fault happened; only switch S7, S8 flown past by fault over-current which
meanwhile,active based on genetic alforithm and launch of direction is opposite with the assumed positive direction of
the fault location algorithm, through the algorithm can solve themselves.Stresses the value of Ij into fitness function, and
the possible fault lines directly. through 52 generation the mininum of fitness function is 0.5
The flowchart of the fault location algorithm flow and the optimal individual is 7.This individual corresponds
based chart of the genetic algorithm is shown in Fig.2.In the to line g.The result consistents with hypothesis, so genetic
flowchart, the value of Ij is determined by fault over-current algorithm can locate correctly.
information that FTU fed back to the SCADA control center. When the switch S6 has distorted informations for
The code should include the analysis of network topology to some reason , the current signal feed back by FTU change
determine the number of switches and the upstream and the from"1" to "-1", modify Ij value of switch S6, simulate new
downstream lines, which is in order to generate the initial round, The results as shown in the figure .
random population. Since the actual fault is almost single
point of failure, the generation of the initial population can
use completely initialize a single point of failure, which will
help reduce the number of searches of the genetic algorithm
and improve the efficiency of the algorithm. The fitness
value of convergence criterion for the optimal solution are
greater than (M-l)ˈ where M is a large integer and it is
taken to be greater than or equal to twice the number of all
of the switches.[5][6]


Figure 3 The simulation diagram

105
As is shown in Fig.3, when population achieves the
minimum fitness value 3.5 by 59 genetic manipulations, the
optimal individual is 7 corresponding faulty line g, which
corresponds with the assumption, and genetic algorithm is
exactly located. It can be seen that under the circumstance of
information distorting, genetic algorithm can still locate
faulty line accurately. When the other switch information
distorting happens, it’s still located accurately by simulation
analysis. Therefore under the circumstance of single bit
information distorting, genetic algorithm location isn’t
influenced at all. Moreover, two-bit information distorting is
also simulated, population iterations increase, and the result
of genetic algorithm location is accurate, impregnable.

VII. CONCLUSION

This paper creatively put forward assuming each switch


to positive direction, positive direction is nonsense to
network , each switch has its own positive direction.After
determining the switch of the positive direction ,the
switching function and fitness function structure is detailed
introduces .This is the key to the application of genetic
algorithms in distributed power distribution network fault
location, and then use the Matlab genetic algorithm toolbox
to analysis the example of multiple fault, simulation analysis
to prove the effectiveness of the algorithm.

ACKNOWLEDGMENT

The project was supported by Tian Jin Natural Science


Foundation (09JCZDJC23900ˈ10JCZDJC23100), Tian Jin
Science and Technology Support
Program(10ZCECJD43080).

REFERENCE

[1] Yangdong Wen, Xin Wang. “The influence of distribution generation


on distribution network power relay protection”. Relay,
2008,3(1),pp.12-14.
[2] Hongwei Du, Yaming Sun, Hongjing Liu. “Based on genetic
algorithm for distribution network fault location and isolation”.
Power System Technology,2000,24(5),pp.52-55
[3] W. P. Luan,M. R. Irving, J. S. “Daniel. Genetic Algorithm for Supply
Restoration and Optimal Load Shedding in Power System
Distribution Networks”. IEEE Proc Gener Transm Distrib. 2002,
149(2), pp. 145-151.
[4] Y. Kumar, B. Das, J. Sharma. “Genetic Algorithm for Supply
Restoration in Distribution System with Priority Customers”.
International Conference on Probabilistic Methods Applied to Power
Systems, 2006, pp. 1-7.
[5] K. Iwasaki, H. Aoki. “Service Restoration Problem in Distribution
Power System Using Improved GA”. Electrical Engineering in Japan,
2009, 166(1), pp.10-19.
[6] Chengshan Wang, Peng Li. “The development and challenge of
distributed power generation, micro network and intelligent
distribution network”. Automation of Electric Power Systems, 2010,
34(2), pp.10-11.

106

You might also like