“Optimal Reactive Power Dispatch (ORPD) Using Particle Swarm
Optimization (PSO)”
By
Kartik S. Pandya
Associate Professor
Dept of Electrical Engineering,
CSPIT,
Charotar University of Science and Technoloogy-Changa,
INDIA
Email:[Link]@[Link]
Acknowledgements: I am very much thankful to Prof. Ray. Zimmerman (the creator of
MATPOWER software) for permitting me to use this software for executing load flow analysis. I
am also thankful to Mr. Jagatpreet Singh whose PSO codes have been used in this application.
His developed PSO codes are available on MATLAB file exchange.
Important note:
I have solved the optimal reactive power dispatch problem using Particle Swarm Optimization
algorithm for IEEE 30 bus test system. I have used MATPOWER[1] software, version 3.2 for
executing load flow analysis for each particle. As MATPOWER is a system requirement, users
are kindly requested to download it from its website first ([Link]/matpower) and
then in the folder of MATPOWER, please include ORPD_30bus.m file. Users are requested to
define some variables as global in some files of MATPOWER as given in table 1.
Table 1: Variables to be defined as global in MATPOWER files
Sr no. File name.m Variables to be defined as global
1 printpf.m global loss
2 runpf.m global bus, gen, branch (In line no. 114)
In runpf.m file, in “default arguments”, please write casename=’case_ieee30’ (In
line no. 60)
ORPD_30bus.m can be used to simulate reactive power dispatch of IEEE 30 bus test system
only. For executing it for other systems, the users will have to make necessary changes in
the input data.
Introduction:
Optimal reactive power dispatch is a complex optimization problem in which we try to
“optimally” set the values of control variables like reactive power output of generators
(generator bus voltages), tap ratios of transformers and reactive power output of shunt
compensators like capacitors etc. to minimize the total transmission active power losses
while satisfying a given set of constraints [2]. I have used basic Particle Swarm
Optimization (PSO) for minimizing the following objective function (1).
All equations are taken from ref. [2].
s.t.
Equality constraints:
Active power flow balance equations at all buses excluding slack bus
Reactive power flow balance equations at all PQ buses (load buses)
Inequality constraints:
Transformer tap-setting constraint
Power flow limit constraint of each transmission line
Static square penalty function is used to handle inequality constrains. So the Augmented
objective function (fitness function) would be as equ(2)
Particle Swarm Optimization
PSO is a fast, simple and efficient population-based optimization method which was
proposed by Eberhart and Kennedy [3]. Each particle updates its position based upon its
own best position, global best position among particles and its previous velocity vector
according to the following equations:
where,
: The velocity of particle at iteration
: Inertia weight of the particle
: The velocity of particle at iteration
: Positive constants having values between [0, 2.5]
: Randomly generated numbers between [0, 1]
: The best position of the particle obtained based upon its own experience
: Global best position of the particle in the population
: The position of particle at iteration
: The position of particle at iteration
: Constriction factor. It may help insure convergence.
Suitable selection of inertia weight provides good balance between global and local
explorations.
Where, is the value of inertia weight at the beginning of iterations, is the value
of inertia weight at the end of iterations, is the current iteration number and is
the maximum number of iterations.
Step by step procedure for ORPD using PSO for IEEE 30 bus test system
(1) Define control variables (vg1, vg2, vg5, vg8, vg11, vg13, T1, T2, T3, T4, QC3,
QC10 and QC24) within their permissible range, define population size, no of
iteration (=200), assume suitable values of PSO parameters, input the data of 30
bus test system
(2) Take iter=0
(3) Randomly generate the population of particles and their velocities
(4) For each particle run NR load flow to find out losses.
(5) Calculate the fitness function of each particle using equ. (2)
(6) Find out “personal best (Pbest)” of all particles and “global best(Gbest)” particle
from their fitnesses
(7) Iter=iter+1
(8) Calculate the velocity of each particle using equ. (3)and adjust it if its limit gets
violated
(9) Calculate the new position of each particle using equ. (4)
(10)For each particle run NR load flow to find out losses.
(11)Calculate the fitness function of each particle using equ. (2)
(12)For each particle if current fitness(P) is better than Pbest then Pbest=P
(13)Set best of Pbest as Gbest
(14)Go to step no. 7, until max. no of iterations is completed.
(15)Coordinate of Gbest particle gives optimized values of control variables and its
fitness gives minimized value of losses.
Simulation results:
The results are stored in “diary” file. As PSO is a stochastic optimization method, it
gives different optimized results in every simulation. I have considered one of the
best results.
Table 1: Comparison of losses obtained before and after optimization
Sr Losses before Losses after % reduction in
no. optimization optimization losses
1 Active power losses
17.53 16.308 (fxmin) 6.97
(Ploss) MW
2 Reactive power
losses (Qloss) 67.59 64.10 5.16
MVAR
3 b=index of “Gbest” particle
Table 2: Values of control variables after optimization
Bus Control variables Optimized values (xmin)
1 Vg1 (pu) 1.092
2 Vg2 1.073
5 Vg5 1.044
8 Vg8 1.051
11 Vg11 1.090
13 Vg13 1.066
6-9 T1 0.979
6-10 T2 1.015
4-12 T3 1.033
27-28 T4 0.982
3 QC3 (MVAR) 10.788
10 QC10 7.589
24 QC24 11.790
Fig. 1: Convergence characteristic of PSO
Table 3: Selected parameters of PSO
1 Population size 50
2 Acceleration constant (C1, C2) 2.1 and 2.0
3 Constriction factor 0.729
4 Max. and Min. inertia weights 1 and 0.2
5 Max. and Min. velocity of particles 0.003 and -0.003
6 Convergence criterion 200 iterations
Note: User may change the values of these variables and may check their effect on the
value of optimized results.
References:
[1] R.D. Zimmerman et al. ,“MATPOWER: Steady-State Operations, Planning and
Analysis Tools for Power Systems Research and Education,” IEEE Trans. On Power
Systems, vol. 26, no. 1, Feb. 2011, pp. 12-19.
[2] B. Zhao et al., “A Multiagent-Based Particle Swarm Optimization Approach for
Optimal Reactive Power Dispatch”, IEEE Trans. On Power Systems, vol. 20, no. 2, May
2005, pp. 1070-1078.
[3] Kennedy J. and Eberhart R, “Particle swarm optimization”, In Proc. of the IEEE
international conference on Neural networks, 1995, pp. 1942-1948.