You are on page 1of 6

IEEE - 40222

Optimal Phasor Measuring Unit Placement


by Binary Particle Swarm Optimization
Saroj Kumari Dr. Pratima Walde Asif Iqbal Akash Tyagi
Deptt of electrical engineering Deptt of electrical engineering Computer Science Division Electrical Engineering Division
Galgotias University Galgotias University Piro Technologies Pvt. Ltd. Piro Technologies Pvt. Ltd.
Greater Noida, (U.P), India Greater Noida, (U. P), India New Delhi, India New Delhi,India
Sarojk2701@gmail.com pratima.walde@galgotiasuniversity.edu.in asif@pirotechnologies.com akashtyagi0010@gmail.com

Abstract— This work presents a technique for the optimal provides time synchronized measurement of current and
Phasor Measurement Units (PMUs) placement for complete voltage phasors at that bus. A number of PMU already
power network observability with number of PMUs as installed in several utilities around the world because of its
minimum as possible. Due to the high installation cost of various applications discussed in paper [2].
PMUs it is necessary to make the system fully observable with
minimum PMUs. A binary particle swarm optimization One of the most important issues is the cost of the PMU
method (BPSO) is implemented on IEEE standard system
which limits its installation, although an increased
and Puducherry 17 bus system. The BPSO method of
demands in future might bring down its cost. Therefore a
Optimum PMU Placement can therefore be applied to any
power system to make the system fully observable with method is required to find the optimal locations of such
different aspects of the power system. The obtained results synchronized measurement devices, in order to minimize
are compared with other techniques and it is found that number of PMUs and make the system fully observable. A
Adaptive GA, GILP, SA, TS, BSA and the proposed BPSO power system is considered observable when all the states
method was found better. in the system can be determined. A lot of research has been
done in determining the minimum number of PMUs
Index Terms—Phasor measuring unit,Binary particle considering full system observability and finding their
swarm optimization, optimal PMU placement, observability.
optimal location. In [3], [4] & [5], the authors used integer
linear programming to find minimum number and optimal
I. INTRODUCTION locations of PMUs. In [6] graph theoretic method is
Synchronized measurements attracted the great attention to considered for placing PMUs which uses the concept of
the power system engineers. It provides the realization of spanning trees of the power system graph in order to
the real time monitoring, control and protection. determine the optimal locations of PMUs. Simulated
Synchronization can be achieved by real time sampling of annealing is used to solve the programmable
current and voltage waveforms with the help of Global communication constrained PMU placement problem. In
Positioning system (GPS). Basically through [7] and [8] Genetic Algorithm procedure for solving OPP
synchronization voltage phase angle can be directly problem is presented. In [9] Tabu search approach is
measured which was not feasible earlier and also increases proposed which finds optimal number of PMUs by trial and
the speed and accuracy of state estimation. error method and their locations was determined by Tabu
search algorithm.
In power system, existing SCADA and EMS only
facilitates the steady state view of the power network. This approach has been successfully used in many
SCADA measurements are obtained at slower rate and application of power system. BPSO based approach has
does not measure the phase angle of voltage of power been used in this paper as an optimization tool to determine
system. Angle measurement are necessary to find possible the optimal number of PMUs for complete power network
instabilities and allow the switching control to mitigate the observability. It is observed that BPSO successfully
threating situation. It gives improved protection and optimizes the objective function and converges at global
advance control of power system. For the above optima at a rate faster than other methods. The simulations
advantages Phasor Measurement Unit (PMU) device is an are carried out using Binary PSO are listed below:-
advance instrument that provides synchronized
measurement which is most accurate. The PMU at a bus

8th ICCCNT 2017


July 3 -5, 2017, IIT Delhi,
Delhi, India
IEEE - 40222

1. Location of PMU in IEEE test system with and N signifies the number of buses in a power network for
without considering ZIBs. PMU installation, is the installation cost vector at the
2. Location of PMUs in a real system considering bus k, x signifies binary decision vector having element ,
ZIBs only. which decides the placement of PMU on the kth bus, A(x)
is the observability constraint or binary connectivity matrix
II. PMU PLACEMENT PROBLEM whose elements are given as:
FORMULATION
Initially binary connectivity matrix A(x) is created
In a power system, a bus is defined as observable when according to criterion given as follows:
node voltages of a bus can be calculated by the use of other
node voltages and branch currents. The basic rule in PMU 1 =
A(k,l) = 1 ℎ (5)
placement is that, the PMU can measure the voltage phasor
0 ℎ
at the bus on which it is placed as well as at the adjacent
buses [10]. A bus without generation and load is called a b= [1 1 1 1 … … … NB times]
zero injection bus (ZIB). No PMU will be placed at ZIB in
any circumstance rather the voltage phasors can be Due to the fact that, rather only considering the cost of the
measured using KCL. PMU, installation cost of the current transformer and other
equipments should also be considered for each
The PMU placement objective is to find the optimal transmission line connected to the bus. Because, of the
number of PMUs with full system observability. The unavailability of equipment costs and other associated cost
objective function can be defined as: of installation it is assumed the installation cost of one
PMU at a bus together with a connected transmission line
1. Optimal PMU placement without considering installation
to be 1 p.u., defined as the base installation cost. The
cost :
installation cost will be further added by 0.1 p.u. for any
additional equipment connected to that bus. The next
Min ∑ (1)
section gives detail of the BPSO used in this paper.
Subject to A(x) ≥ b (2)

III. BINARY PARTICLE SWARM


2. Optimal PMU placement considering installation cost : OPTIMIZATION

To find the total installation cost of PMU, an installation The proposed problem is mathematically formulated in
cost factor ‘c’ is defined in this paper, its elements gives the previous sections and it was discussed that Meta
the installation cost of PMUs in each bus of the power Heuristic technique has been very efficient with solving
system network. Optimal PMU placement problem Optimal PMU Placement (OPP) problem and on this basis
considering the installation cost criterion is given as we decided to apply Binary form of one of the popular
follows: Meta heuristic technique called Particle swarm
Optimization (PSO) technique.
Min ∑ (3)
Advantage of Particle swarm Optimization (PSO) is that it
Subject to the same constraint given in eq. (2) has easy implementation, more efficient have control
parameters and have fewer parameters to adjust. However
Where,
the drawback of this technique is higher computation time
Determines the feasibility installation of PMU at the ith and that also increases with the increasing the dimensions
bus defined as follows: of solution i.e. number of buses and complexity of
constraints. Even though this problem is an offline
1 planning problem so we still propose BPSO method that
= (4)
0 ℎ simultaneously make sure that the objective solution are

8th ICCCNT 2017


July 3 -5, 2017, IIT Delhi,
Delhi, India
IEEE - 40222

within feasible region i.e. constraints are satisfied and also , =f( ) (10)
convergence is accelerated.
At every iteration, velocity and position of each particle are
updated based on precious best position(P ) and best
position calculated by information(G ). An updated
A. Particle Swarm Optimization formula is given as:
Particle Swarm Optimization (PSO) was first introduced ( )
= + P − + G −
by Kennedy and Eberhart in 1995. It is an evolutionary , , , ,
( )
computation technique which is inspired by the social , (11)
behavior of group of organisms like bird flocking.
, = , + , (12)
It is a population search based method where individuals
called as particles that adjust their positions in search space Where i is index of the particle, j is the index of the position
with the given velocities and direction. In this swarm of of the particles in the swarm, k is the iteration number.
birds that don’t have leader, so they find food by random is the inertia parameter and its ranges is in between 0.95 to
movement and follow one of the member of group whose 0.99. called cognitive parameter and has value around
position is closest with food source. The group achieves
2.0 called social parameter and its value equals to .
their best condition simultaneously by communicating with
Parameters , are random numbers consistently
the members who are already at better solution. So they
distributed between 0 and 1.
will move simultaneously to best position. This would
repeatedly happen until best food source/ condition has B. Binary Particle Swarm Optimization
discovered.
The BPSO technique was first introduced by Kennedy and
It’s an iterative method in which every particle adjusts its Eberhart in 1997. In BPSO each element of position vector
velocity and position for best solution. Technically the takes only binary values (0 or 1). Here velocity remains
commutative cognitive occurrence of the swarm turns out unaltered but position is updated and rewritten by rule:
into the optimization technique. Mathematically let us
suppose we have N particles within the swarm in the = + P −
( )
, , , ,
problem space. Each particle has position and velocity that ( )
directs flying of particles and also has cost value that to be + G − ,

minimized.
( )
0, () ≥ ( , )
Position and velocity of the ith particle in jth dimension at , = ( )
(13)
1, () < ( , )
kth iteration are denoted by x , and v , :
Where S is sigmoid function to transfer the velocity to the
, =⌊ ⋯⋯ ⌋ (6) possibility defined as:

, =⌊ ⋯⋯ ⌋ (7) ( )
( )= , (14)
Where j=1,2,3,………., up to n dimension
And rand () is a random number with range [0, 1].
Swarm maintains its global best position:
Here, it is shown that the BPSO successfully optimizes the
=⌊ ⋯⋯ ⌋ (8) given test function with a faster rate of convergence at
,
global optima.
=f( ) (9)
The updated velocity and the acceleration coefficients
And each particle maintains its individual best position: , decide the effect of P and G on the particles
current velocity vector. Better convergence can be obtained

8th ICCCNT 2017


July 3 -5, 2017, IIT Delhi,
Delhi, India
IEEE - 40222

by making acceleration constant and inertia constant


dependent one each other as shown below:
Test system No. of No. of ZIB Location of ZIB

= (15) branches
∅ ∅
IEEE 14-bus 20 1 7
system
= =∅ (16)
Indian Utility 21 2 1 4
IV. CASE STUDIES (Pondicherry)
17 bus
The proposed BPSO method is tested on IEEE 14 bus system
system [14] and its effectiveness is evaluated by applying
the proposed approach on the Indian Utility-Puducherry-17 TABLE I
bus test system [15]. Software has been developed using BASIC CONFIGURATION OF THE TEST SYSTEMS
MATLAB to test the suggested BPSO technique and
simulations are carried out on a PC having Intel Core i3
Processor@2.20 GHz, 2 GB RAM. Also the presence of
ZIBs in PMU placement minimizes the required number of
PMUs for entire power network observability. Both the
cases with and without zero injection bus have been
performed. Single line diagram of the IEEE 14 bus system
and 17 bus system is shown in Fig. 1 and Fig. 2
respectively. Table 1 shows the number of zero injection
buses for the two test systems. Table 2 shows the
optimization parameters for the BPSO used in OPP
problem. These values of optimization parameters are
chosen after several runs of the program and gives the best
converging solution.

Number of buses defines the problem space dimension. To


achieve better search in the problem space, the BPSO
program is run number of times for each power system
which is mention in Table III and table IV for IEEE 14 bus
system and Puducherry 17 bus system. Fig. 1. IEEE 14 bus test system [14]

Table III shows several alternative PMU locations for 14


bus system that result in full system observability but only 1

one out of several solutions will gives the minimum


installation cost of 4.6 p.u. Similarly for 17 bus system in
2
3

table IV, minimum installation cost is 7.1 p.u. The


program was run 100 of times for the two systems. In Table 4
5

15 11

V, according to the result it can be seen that by equipping 7

three PMUs in the 14 bus system, full observability will be 16

achieved at minimum installation cost. Similarly for 17 bus 14

system minimum six PMUs required at the defined location 6

to get full system observability at minimum installation 12

cost. The best achieved solution including PMU locations 8

along with installation cost are listed in table V and VI. 10 13

9 17

Fig. 2. 17 bus system [15]

8th ICCCNT 2017


July 3 -5, 2017, IIT Delhi,
Delhi, India
IEEE - 40222

TABLE II TABLE VI
BPSO Optimization parameters OPTIMAL NUMBER OF PMUs FROM BPSO FOR 17 BUS SYSTEM
Parameter names Values
Methods No. of PMU Locations Cost
Population size 20 PMUs
Cognitive parameter( ) 2.04 Excluding cost 6 1 4 5 8 11 17 7.4
factor
Social parameter ( ) 2.04 (proposed)
Inertia parameter( ) 0.95 - .099 Including cost 6 1 4 5 9 16 17 7.1
factor
Random parameter ( and ) 0–1 (proposed)
Considering ZI 6 2 8 11 12 14 6.7
Bus 17

TABLE III
OPTIMUM PMU PLACEMENT CONSIDERING INSTALLATION TABLE VII
COST FOR 14 BUS SYSTEM OPTIMAL NUMBER OF PMUs RESULTING FROM BPSO AND
OTHER OPTIMIZATION TECHNIQUES CONSIDERING ZI BUSES
No. of Optimal PMU No. of Installation No. of
trials location PMUs Cost (p.u.) Buses
Observed Test Systems Proposed Method
1 2 7 10 13 4 4.8 14 No. of Location of Installation
2 2 7 11 13 4 4.8 14 PMUs PMU Cost
3 2 8 10 13 4 4.6 14 IEEE 14 bus 3 269 3.9
4 1 3 8 10 13 5 5.5 14 system

5 2 6 8 9 4 4.9 14 Indian utility 6 2 8 11 12 14 17 6.7


17 bus
systems
TABLE IV
OPTIMUM PMU PLACEMNT CONSIDERING
INSTALLATION COST FOR 14 BUS SYSTEM
Test Graph Three SA and BSA GA
No. of Optimal PMU No. of Installation No. of Systems theoretic stage TS [5] [6] [7]
trials location PMUs Cost (p.u.) Buses
procedure method
Observed
1 1 4 5 9 16 17 6 7.1 17 [13] [4]

2 2 7 8 11 12 14 7 7.7 17 IEEE 14 5 3 3 3 3
17 bus
3 1 4 5 10 13 16 6 7.3 17
system
4 2 4 5 8 16 17 6 7.3 17
Indian NR* NR* NR* NR* NR*
5 1 4 5 9 16 17 6 7.2 17
utility
17 bus
system
TABLE V
NUMBER OF PMUs RESULTING FROM BPSO FOR 14 BUS NR* means Not Reported
SYSTEM

Methods No. of PMUs PMU Locations Cost


Excluding cost 4 2 6 8 9 4.9
factor (proposed)
Including cost 4 2 8 10 13 4.6
factor (proposed)
Considering ZIB 3 2 6 9 3.9

8th ICCCNT 2017


July 3 -5, 2017, IIT Delhi,
Delhi, India
IEEE - 40222

VI. REFERENCES
[1] Chakarbarti S, Kyriakides E, “Optimal placement of phasor
measurement units for power system observability,” IEEE Transaction on
Power Systems, vol. 23, No. 3, August 2008.

[2] Ree Jaime, Centeno V, Thorp J, Phadke, “Synchronized phasor


measurement applications in power system,” IEEE Transaction on Smart
Grid, vol. No. 1, June 2010.

[3] A Azizi S, Salehi Dobakhshari A, Nezam Sarmadi SA, Ranjbar AM,


Gharehpetian GB, “Optimal multi-stage PMU placement in electric power
systems using Boolean algebra,” Int Trans Electr Energy Syst 2014;
24(4):562–77.

[4] C. Jian and A. Abur, "Placement of PMUs to Enable Bad Data


Fig. 3.Convergence rate of BPSO for 14 bus system Detection in State Estimation," Power Systems, IEEE Transactions on,
vol. 21, pp. 1608-1615, 2006.

[5] Gou B, “Optimal placement of PMUs by integer linear programming,”


IEEE Transactions on Power Systems, Vol. 23, No. 3, August 2008

[6] Nuqui R, Phadke A,” Phasor measurement unit placement techniques


for complete and incomplete observability,” IEEE Transaction on power
delivery, vol. 20, No. 4, October 2005.

[7] Abiri E, Rashidi F, Niknam T,”An optimal PMU placement method


for power system observability under various contingencies,” Int Trans
Electr Energy Syst 2015; 25(4):589–606.

[8] Marin FJ, Garcia-Lagos F, Joya G, Sandoval F, “ Genetic algorithms


for optimal of SID placement of phasor measurement units in electrical
networks,” Electronics Letters September 2003; 39(19):1403–1405

Fig. 4. Convergence rate of BPSO for 17 bus system [9] Amany El Zonkoly et al, ”Optimal placement of PMUs using
improved tabu search for complete observability and out-of-step
prediction,” Turkish Journal of Electrical Engineering & Computer
Sciences (2013) 21: 1376 -1393
V. CONCLUSION
[10] Chakrabarti S, Venayagamoorthy G, Kyriakides E, “PMU placement
A BPSO based methodology is proposed to find the
for power system observability using binary particle swarm optimization.
optimal location of PMUs in a network for its entire 2008
observability at minimum installation cost. The proposed
method have higher rate of convergence to the global [11] Abiri E, Rashidi F, Niknam T. An optimal PMU placement method
for power system observability under various contingencies,” Int Trans
solution with less execution time when compared to the
Electr Energy Syst 2015; 25(4):589–606
existing methods. The PMU placement problem has been
associated with different aspect with the presence of ZIBs. [12] T.L. Baldwin, L. Mili, M. B. Boisen, R. Adapa, “Power system
Simulation results for the different power networks shows observability with minimum phasor measurement placement,” IEEE
Trans. Power System, vol. 20, pp. 707-715, May. 1993.
the efficiency and the robustness of the proposed method
in achieving the complete network observability with [13] B.Pal and B.Chaudhuri, “Robust control in Power systems,” Springer
minimum number of PMUs at minimum installation cost. Verlag, London, 2005.
This iterative process overcomes the limitations of
[14] Power system test case archive
conventional optimization method such as integer linear http://www.ee.washington.edu/research/pstca/. Last Accessed on 15-04-
programming. And also its advantage of simplicity, 2017 23:31.
scalability to large systems, computation efficiency.
[15] Raj P A. “Performance evaluation of swarm intelligence based power
Future work will include additional constraints such as
system optimization strategies,” Phd, Department of Electronics and
measurement uncertainty and redundancy at the buses. Communication Engineering, Pondicherry, 2008

8th ICCCNT 2017


July 3 -5, 2017, IIT Delhi,
Delhi, India

You might also like