Vous êtes sur la page 1sur 4

International Conference on Power, Energy and Innovation (ICPEI 2019)

October 16-18, 2019, Pattaya, THAILAND

Real and Reactive Powers Decomposition


Optimal Power Flow Using Particle
Swarm Optimization
Kanatip Rojanaworahiran Keerati Chayakulkheeree
School of Electrical Engineering,
Institute of Engineering, Suranaree University of Technology
Nakhonratchasima, Thailand
E-mail : kanatip_mang@hotmail.com, keerati.ch@sut.ac.th

Abstract— This paper proposes a real and reactive iteratively. In the proposed real and reactive powers
powers decomposition optimal power flow decomposition optimal power flow (PQDOPF), the
(PQDOPF). The problem formulation was solved by problem formulation is decoupled into the total cost
particle swarm optimization (PSO). In the PQDOPF minimization subproblem (TCMS) and the total real
problem formulation, the total cost minimization power loss minimization subproblem (TLMS). In the
subproblem (TCMS) and the total real power loss TCMS, the total generation cost minimization problem is
minimization subproblem (TLMS) are solved by PSO solved by PSO. The optimal real power generation of
successively. The PQDOPF algorithm had been each bus is the output. Meanwhile, in the TRMS, the
simulated with IEEE 30 buses system. The simulation total real power loss minimization problem is determined
result shown that the proposed algorithm can by PSO and the optimal generator voltage magnitudes
efficiently minimize the total operating cost with total and transformer tap changing positions are the outputs.
system loss consideration. The TCMS and TRMS are solved sequentially, for the
lowest total generation cost of the system. The
Keywords— Particle Swarm Optimization, Optimal simulation result with IEEE 30 buses system shown that
Power Flow, Total cost minimization, Total real power the proposed algorithm can resourcefully minimize the
loss minimization total cost and total real power loss, comparing to the
earlier method.
I. INTRODUCTION The organization of this paper is as follows. Section
Optimal power flow (OPF) is an important short II. addresses the PQDOPF problem formulation. PSO for
term analysis tool for optimal operation of electricity solving the PQDOPF is given in Section III. The
utility, to meet the system load, at the lowest possible simulation result on the IEEE 30 bus system are
cost, subject to transmission and operational constraints. illustrated in Section IV. Lastly, the conclusion is given
Now a day, the OPF problem is solved by computer in Section V.
software with high difficulty computation.
In typical, the OPF model represents the problem of
determining the best operating levels for electric power II. PQDOPF PROBLEM FORMULATION
plants in order to meet demands given throughout a In the PQDOPF problem formulation, the OPF
transmission network, with the objective of minimizing objective function can be expressed as multi-objective
operating cost. Many methods have been proposed such problem which are; (i) total cost minimization
as Genetic Algorithm (GA) [1], Particle Swarm subproblem (TCMS) and (ii) total real power loss
Optimization (PSO) [2-4], Ant Colony Optimization minimization subproblem (TLMS). The optimization
(ACO) [5] and Tabu Search (TS) [6] in OPF, to obtain problem formulation is as follow,
the power system optimal result. Among these stochastic
(i) minimize total operating cost in TCMS as,
optimization methods, PSO is a famous stochastic base
optimization technique developed by Dr. Eberhart and NG
minimize FC = F ( P ) , (1)
Dr. Kennedy in 1995 [7], inspired by social of behavior  Gi
of bird flocking or fish schooling. PSO was proven to be i =1

one of the best stochastic optimization method for and


several problems. (ii) minimize total real power loss in TLMS as,
Due to the weak linkage nature between real and
reactive power in power system, in this paper, the real minimize NB NB
, (2)
PL =  Gij Vi 2 + V j2 − 2VV 
i j cos(δ i − δ j ) 
and reactive power flow optimization problem are solved i =1 j =1
j ≠i
International Conference on Power, Energy and Innovation (ICPEI 2019)
October 16-18, 2019, Pattaya, THAILAND

subject to the power balance constraints, III. PSO BASED PQDOPF


NB In a PSO system, the initialized group of random
PG i − PDi =  | Vi || V j || yij | cos(θ ij − δ ij ), i = 1,..., NB , (3) particles are generated and then the objective function
j =1
for each particle is computed. Afterward, the best
NB
QG i − QDi = − | Vi || V j || yij | sin(θ ij − δ ij ), i = 1,.., NB , (4) solution is obtained and then the particles are updated in
j =1 the next generation. In every iteration, each particle is
updated by following the best values. The first one is the
and line flow limit and transformer loading constraints, best solution (fitness) it has achieved so far. This value is
| f i |≤ f i max ,for i = 1,…NL, (5) called pbest. Another value that is tracked by the particle
swarm optimization algorithm, which is the best value
obtained so far by any particle in the population. This
and real power generation constraint, best value is a global best and called gbest. When a
PGimin ≤ PGi ≤ PGimax , for i = 1,…,NL, (6) particle takes part of the population as its topological
neighbors, the best value is a local best and is called
and reactive power generation constraints, lbest. The PSO computational step can be given as
QGimin ≤ QGi ≤ QGimax , for i = 1,…,NL, (7) follow,
Step 1: Set the iteration counter and create a
and bus voltage limit constraint, ‘population’ of agents (particles) uniformly
Vi min ≤ Vi ≤ Vi max , for i = 1,…,NL, (8) distributed over searching space. In TCMS, the
particles are PGi. In TLMS, the particles are |Vi|
at the generator buses and Ti.
and transformer tap limit constraint, Step 2: Evaluate each particle’s position according to
Ti min ≤ Ti ≤ Ti max , for i = 1,…,NL. (9) the objective function, which are Eq.(1) for
TCMS and Eq.(2) for TLMS.
Where, Step 3: If a particle’s current position is better than its
previous best position, then update the particle.
FC is the total system operating cost ($/hr.), Step 4: Determine the best particle (according to the
F(PGi) is the operating cost of the generator connected particle’s previous best position).
at bus i ($/hr.), Step 5: Update particles velocities by Eq. (10).
fimax is the limit of line flow or transformer (MVA), Step 6: Move particles to their new position by Eq.
fi is the MVA flow of line or transformer i (11).
(MVA), Step 7: Evaluate each particle’s position according to
Gij is the conductance of the lines between bus i
and bus j for j≠i, the objective function, which are TCMS or
NG is total number of generators, TLMS
NB is the total number of buses,
NT is the total number of transformers, Step 8: Does the objective function value (TCMS or
PL is total real power loss (MW), TLMS) is satisfied? or Does the iteration reach
PDi is real power at bus i (MW), the maximum number of iterations? If yes, go to
PGi is the real of power generator connected bus i step 9. If no, go to step3
(MW), Step 9: Stop.
PGimax is the maximum real power generation at bus i
The particles velocities and updating can be
(MW),
expressed as,
QGi is the reactive power generator at bus i
(MVAR), vit+1 = wvit + c1r1(pbestit - pit) + c2r2(gbestt - pit), (10)
QDi is the reactive power demand at bus i (MVAR),
|Vi| is the voltage magnitude of bus i (p.u.), xit+1 = xit + vit+1 , (11)
Vi is the voltage of bus i (p.u.), where,
|yij| is the magnitude of the yij element of Ybus w is the inertia weight factor = 1,
(mho), c1,c2 are the acceleration constants in the range [0,2],
θij is the angle of the yij element of Ybus (radian), r1,r2 are the uniform random values in the range
and [0,1],
δ ij is the voltage angle difference between bus i t is the number of iteration,
v is the velocity of particle,
and j (radian).
x is the position of particle,
pbest is the best particle position,
gbest is the best group position, and
p is the particle.
International Conference on Power, Energy and Innovation (ICPEI 2019)
October 16-18, 2019, Pattaya, THAILAND

TABLE I. THE COMPARISON RESULTS FOR IEEE 30 BUSES

Proposed
Variables Base case [8] [1]
PQDOPF
V1 1.050 1.050 1.050 1.070
V2 1.045 1.045 1.038 1.050
V5 1.010 1.011 1.012 1.021
V8 1.010 1.019 1.012 1.032
V11 1.050 1.092 1.082 1.088
V13 1.050 1.091 1.067 1.075
T1 0.978 1.028 1.013 0.989
T2 0.969 0.960 0.950 0.977
T3 0.932 1.0047 1.000 0.942
T4 0.968 0.9416 0.9625 0.975
PG1 93.4 167.66 176.20 176.601
PG2 80 48.84 48.75 48.607
PG5 50 21.51 21.44 21.483
PG8 20 22.15 22.95 21.750
PG11 20 12.14 12.42 12.077
PG13 20 12.00 12.02 12.00
Total Cost
900.76 802.400 802.060 800.985
($/hr.)
Total
losses 18.742 9.642 9.380 9.118
(MW)

TABLE II. THE OPTIMAL BUS VOLTAGE MAGNITUDE AND ANGLE


RESULT FROM PQDOPF FOR IEEE 30 BUSES SYSTEM
Fig 1. PQDOPF Computational Procedure
Bus number Voltage (p.u.) Phase angle (Degree)
1 1.070 0.000
In the TCMS, the real power generation of generator 2 1.050 -5.193
(PGi) are the output solved by PSO. Meanwhile, in the 3 1.033 -7.413
TLMS, the generator bus voltage magnitudes (|Vi|) and 4 1.024 -9.134
transformer tap positions (Ti) are the output obtained by 5 1.021 -13.889
PSO. The propose PSO based PQDOPF 6 1.023 -10.896
7 1.015 -12.640
computational procedure can be illustrated as in Fig1.
8 1.022 -11.605
9 1.054 -13.936
10 1.047 -15.525
IV. SIMULATION RESULTS
11 1.088 -13.936
The PQDOPF has been verified on the IEEE 30 12 1.060 -14.744
buses shown in Fig 2. The system line data and bus data 13 1.075 -14.744
were obtained from [8]. The results of PQDOPF, 14 1.046 -15.620
15 1.041 -15.705
comparing to the methods in the previous works, are 16 1.047 -15.345
shown in Table 1-3. 17 1.042 -15.667
18 1.031 -16.332
19 1.028 -16.514
20 1.032 -16.324
21 1.032 -16.035
22 1.038 -15.861
23 1.032 -16.042
24 1.027 -16.226
25 1.021 -15.837
26 1.006 -16.251
27 1.031 -16.251
28 1.022 -11.542
29 1.011 -16.551
30 1.000 -17.421

In table I, the base case power flow, the methods


propose by [8] and [1], and the PQDOPF are addressed.
The result shown that the PQDOPF provide the
minimum total cost and total loss among all methods,
with the total cost of 800.985 $/hr. and the total loss of
9.118 MW. The PQDOPF solution for voltage
Fig 2. IEEE 30 Buses Test System magnitude and angle and the optimal real power
International Conference on Power, Energy and Innovation (ICPEI 2019)
October 16-18, 2019, Pattaya, THAILAND

generation schedule are shown in Table II and III, V. CONCLUSION


respectively. The transmission line flow and transformer In this paper, the PQDOPF is proposed and
loading resulted from the PQDOPF are within limits as investigated. The total system operating cost and the
addressed in Table IV. It is confirmed that the power total real power loss can be minimized successively. The
flow solution resulted by the PQDOPF is feasible and results shown that the PQDOPF can efficiently solve for
the constraints are within the limits. the optimal solution of the control variables which are
real power generations, voltage magnitudes, and
TABLE III. THE OPTIMAL REAL POWER GENERATION, TOTAL COST,
AND TOTAL REAL POWER LOSS RESULTED FROM PQDOPF OF IEEE 30 transformer tap changings, satisfying transmission and
BUS SYSTEM system operation constraints.
Power Power Total Total
Generator MW Generator Cost Loss
At bus limit (MW) ($/hr) (MW) REFFERENCES
1 176.601 50≤ Pg ≤200 [1] Anastasios G. Bakirtzis, Pandel N. Biskas,
2 48.607 20≤ Pg ≤80 Christoforos E. Zoumas and Vasilios Petrids,
5 21.483 15≤ Pg ≤50 800.985 9.118
8 21.750 10≤ Pg ≤35
“Optimal Power Flow by Enhanced Genetic
11 12.077 10≤ Pg ≤30 Algorithm”, IEEE Transactions on power systems,
13 12.00 12≤ Pg ≤40 vol.17, no2, pp. 229-236, May 2002.
Power demand = 283.4 MW [2] Djilani BEN ATTOUS, Yacine LABBI, “Particle
Swarm Optimization based Optimal Power Flow for
TABLE IV. THE TRANSMISSION LINE FLOW AND TRANSFORMER units with Smooth Fuel Cost Functions”, 2009
LOADING OF THE PQDOPF SOLUTION FOR IEEE 30 BUSES SYSTEM
International Conference on Electrical and
From Bus To Bus Line Flow Line Flow Limit Electronics Engineering-ELECO-2009, International
(MVA) (MVA) Conference, pp. 377-381, 5-8 Nov 2009.
1 2 118.010 130 [3] N. Tejes wara Rao, Jagannath Ch Yadav B., Anyapu
1 3 59.136 130
Jagannadham, “Optimal Reactive Power flow
2 4 34.289 65
3 4 55.068 130 Control For Minimization Of Active Power Losses
2 5 63.370 130 Using Particle Swarm Optimization”, 2015
2 6 45.216 65 Conference on Power,Control, Communication and
4 6 49.829 90 Computational Technologies for Sustainable Growth
5 7 12.604 70 (PCCCTSG), Kurnoolm Andhra Pradesh, India, 1 1 -
6 7 34.669 130
6 8 11.249 32
12 Dec 2015.
6 9 22.976 65 [4] K. Chayakulkheeree and W. Kamklar, “Network
6 10 13.576 32 and Generator Constrained Economic Dispatch
9 11 18.942 65 Using Hybrid PSO-QP Algorithm,” ICTI
9 10 31.329 65 Conference 2006.
4 12 35.144 65
12 13 14.302 65
[5] J.Soares, T. Sousa, Z.A. Vala, H morais, P. Faria,
12 14 8.2475 32 “Ant Colony Search Algorithm for Optimal Power
12 15 19.308 32 Flow Problem”, 2011 IEEE Power and Energy
12 16 7.9761 32 Society General Meeting, 24-29 July 2011.
14 15 1.7353 16 [6] Chunjie Li, Huiru Zhao, Tao Chen, “The hybrid
16 17 3.9871 16
differential evolution algorithm for optimal power
15 18 6.267 16
18 19 2.8854 16 flow based on simulated annealing and tabu search”,
19 20 7.250 32 2010 International Conference on Management and
10 20 9.721 32 Service Science, 24-26, Aug 2010.
10 17 7.048 32 [7] J. Kennedy and R. Eberthart, “Particle Swarm
10 21 18.692 32 Optimization”, Proc.IEEE Int. Conf Neural
10 22 8.905 32
21 22 2.413 32
Networks, vol IV, pp.1942-1948, 27 Nov-1 Dec
15 23 5.9164 16 1995.
22 24 6.600 16 [8] O. Alsac and B. Stott, “Optimal load flow with
23 24 2.285 16 steady state security”, IEEE trans, Power Apparatus
24 25 0.9433 16 and System,. vol.PAS93, no. 3, pp.745-751, May
25 26 4.260 16
1974
25 27 4.3359 16
28 27 17.364 65
27 29 6.406 16
27 30 7.279 16
29 30 3.752 16
8 28 2.276 32
6 28 15.362 32

Vous aimerez peut-être aussi