You are on page 1of 6

Proceedings of the 4th International Conference on ThAC.

5
Systems and Control, Sousse, Tunisia, April 28-30, 2015

Optimal Control of Frequency and Voltage Variations Using PID


Controller Based on Particle Swarm Optimization

Nour EL Yakine Kouba, Mohamed Menaa, Mourad Hasni and Mohamed Boudour

Abstract— This paper deals with an optimal tuning of the with the load that is changing continually, and the change in
Proportional-Integral-Derivation (PID) controller for both real power affect the system frequency, while reactive
Load Frequency Control (LFC) and Automatic Voltage power is less sensitive to the change in frequency and is
Regulator (AVR) of two-area interconnected power system more dependent on changes in voltage magnitude [3-6]. In
using Particle Swarm Optimization (PSO) algorithm. The order to maintain the frequency and voltage as constant, the
active and reactive powers are controlled separately. The LFC
quality of power generation must respect certain minimum
loop controls the frequency and active power and the AVR
loop adjusts the voltage and reactive power. In order to standards. The speed governor is to adjust the frequency and
analyze the system frequency, the tie line power flow, and the real power and hold their values at the scheduled values. In
system voltage, the two-area interconnected power system is other hand each generator is equipped with an excitation
simulated for a step load disturbance in Area-1. The main control to regulate the voltage magnitude and reactive
primary objective is to suppress all the fluctuations of the power at the nominal values.
system due to the disturbance and get back the frequency and Many control schemes such as the conventional
voltage at nominal values. The results are compared with the proportional integral (PI) controller and proportional-
ones obtained by the traditional Ziegler-Nichols (Z-N) method, integral-derivation (PID) controller and optimal control
Genetics Algorithm (GA) and Bacterial Foraging Optimization
Algorithm (BFOA), and the proposed method has proven to be
have been proposed to achieve improved performance [6-8].
very efficient. In 1942, Ziegler and Nichols proposed two heuristic
approaches based on their experience and some simulations
Keywords-Load Frequency Control (LFC); Automatic to quickly adjust the controller parameters P, PI and PID
Voltage Regulator (AVR); Automatic Generation Control [9]. In early 1970 Fosha and Elgerd in their pioneering work
(AGC); PID Controller; Particle Swarm Optimization (PSO). applied classical optimal control methodology to solve LFC
problems [7]. In the 1990s in order to provide simple rules
I. INTRODUCTION
but more efficient than those of Ziegler-Nichols, Åström et
In recent years, power system stability has been al analyzed the adjustment dynamics of a large number of
recognized as an important problem. In modern power process behavior. This analysis led to the establishment of
system, the automatic generation control (AGC) (LFC and tables used in the calculation of P, I and D from simple
AVR) is becoming more significant [1]. In an measurements [7-10]. In 2004 Zwe-Lee Gaing has
interconnected power system each generator is equipped presented PSO for optimum design of PID controller in
with a load frequency control (speed governor) and AVR system [6-11].
automatic voltage regulator (excitation system) to control In order for tuning the optimal values for the PID
the frequency and the voltage magnitude respectively [2-3]. controller parameters, we propose to use the heuristic
Thus, the cross-coupling between the load frequency control Particle Swarm Optimization (PSO) method. PSO is a
(LFC) and the automatic voltage regulator is negligible stochastic search technique developed by Kennedy and
because the speed governor constant time is much bigger Eberhart in 1995 [12], which has been found to be robust
than the excitation time constant, and the transient of the and flexible in solving optimization problem, because it can
excitation system does not affect the LFC dynamic, for this generate a high-quality solution within shorter calculation
reason the load frequency and excitation voltage are time and stable convergence characteristic than other
analyzed independently [4-5]. Over the last century power stochastic methods [13-14].
systems have evolved significantly, and the importance of In this paper, the coupling effects between the load
electricity grew along with the complexity of power frequency controller (LFC) and the automatic voltage
systems. Thus, the different component and power plants regulator (AVR) in two-area interconnected power system
connected to the power network are sensitive to the using optimal PID controller scheme based on Particle
continuity and quality of power supply such as frequency Swarm Optimization (PSO) algorithm are studied. The main
and voltage [6]. The frequency has an inverse relationship idea of this paper is to show the mutual interaction effects
between the combination of both LFC and AVR control
loops as an AGC control system. The interconnected system
_________________________________
Nour EL Yakine Kouba, Mohamed Menaa, Mourad Hasni, and Mohamed is simulated for 0.1pu step load disturbance in area-1 and
Boudour are with the laboratory of electrical and industrial systems, the proposed approach is compared to the classical method
University of Sciences and Technology Houari Boumediene, Bab Ezzouar, of Ziegler-Nichols, Genetics algorithm (GA) and Bacterial
Algiers, Algeria. (E-mail: nkouba@usthb.dz; mmenaa@usthb.dz; Foraging optimization algorithm (BFOA).
mhasni@usthb.dz; mboudour@ieee.org).

978-1-4799-8318-6/15/$31.00 ©2015 IEEE 424


II. INTERCONNECTED POWER SYSTEM MODEL B. Generator Model
The interconnected two-area power system model used The generator model used in this paper is shown in Fig.3
in this paper is shown in Fig.1. This model is composed of [16].
two plants; each plant is composed of a governor, turbine,
load frequency control (LFC), automatic voltage regulator
(AVR), power system stabilizer (PSS) and generator, and
the model of such plant is shown in Fig.2. LFC and AVR
signals are inputted to each unit.

¨P12
Plant 2 Plant 1
¨f2 ¨f1

Area-2 Area-1
Fig.1. Two-area interconnected power system.

A. Power Plant Controls Loops Model


In the two-area interconnected power system each plant Fig.3. Block diagram representation of generator model.
is equipped with governor and excitation system to regulate
the frequency and the voltage magnitude. Also each The mathematical equations system of such model is
generator is equipped with a power system stabilizer (PSS) given by [15-17]:
which is designed for suppressing low-frequency
d' w
oscillations in power systems. Fig.2 (a) graphically depicts M 'Pm  'Pe (1)
dt
the AGC (LFC/AVR) mechanism and Fig.2 (b) shown the
power plant model realized in Simulink/MATLAB. 'Pe Ps.'G  K 2.E ' q (2)
'Vt K 5.'G  K 6.E ' q (3)

dE' q 1
.( KG.(Vf  K 4.'G )  E'q ) (4)
dt Tg
C. Governor Control System Model
To bring the frequency back to the nominal value each
(a) [15] generator with governor adjusts the turbine valve/gate (self
regulation). The schematics of such governor control system
that we used in this work are shown in Fig. 4 [1].

Fig.4. Block diagram of governor-turbine model.

The mathematical formulations of this model are given by:


dPgv 1
SC  KG * dw  Pgv (5)
dt Tsr

dPT 1
Pgv  PT (6)
(b) dt Tsc

Fig.2. Controls loops of a synchronous generator. D. Load Frequency Control (LFC) Model
The main primary objective of LFC loop is to adjust
The main objective of installed LFC and AVR operating point reference of governor units in the control
controllers is to observe the system and take care of small area and to set their outputs. To evaluate the area
changes in load demand to hold the system frequency and requirement (AR) the actual frequency and net interchange
voltage in the nominal values. Small changes in the real power flow are measured by the independent system
power depend on changes in power angle į and the operator (ISO). The area control error (ACE) is given by [1-
frequency f [3-6]. In this paper we use a simple governor- 4]:
turbine model, an excitation system type DC1, and a
simplified linear model of a synchronous machine. ACEi 'PTij  Ei 'wi (7)

978-1-4799-8318-6/15/$31.00 ©2015 IEEE 425


In order to analyze the system frequency, the LFC model G. Tie-Line Model
shown in Fig.5 is used in this research. Using DC load flow method, and assuming that the tie-
line power flow is from area-1 to area-2. The deviation ¨PT12
from the nominal flow can be expressed as follow [17]:
T
'PT 12 ('Z1  'Z 2 ) (13)
S
III. PARTICLE SWARM OPTIMIZATION TECHNIQUE
Fig.5. Block diagram of LFC model. In this paper we use the error function Integral Time of
Absolute errors (ITAE) as an objective function (the fitness
E. Automatic Voltage Regulator (AVR) Model function for PSO). The main objective of this paper is to use
To hold the voltage magnitude constant, each generator the PSO algorithm in order to minimize the error function
is equipped with an excitation system (AVR). The AVR ITAE given in Eq. (15) and obtain the optimal PID
model used in this paper is shown in Fig.6 [18]. controller.
The ITAE performance criterion formula is as follows:
f f
ITAE ³ t. x t  y t .dt ³ t. e t .dt (14)
0 0
t
ITAE ³ t.( 'w1  'w2  ' P12  'V1  'V2 ).dt (15)
0

The particle swarm optimization (PSO) is a heuristic


optimization method based on swarm intelligence. It comes
from research on the bird and fish flock movement behavior
Fig.6. Block diagram of AVR model. [10]. PSO is a population-based optimization method
developed in 1995 by Dr. Kennedy and Dr.Eberhart [12],
The mathematical equations of AVR are given by: and it has become one of the most popular techniques
applied in various optimization problems due to its easiness
Ve Vref  Rf (8)
and capability to find near optimal solutions. It belongs to
dVR 1 the class of direct search methods that can be used to find a
( KA.Ve  VR) (9) solution to an optimization problem in a search space. In the
dt TA
PSO method, a swarm consists of a set of individuals, with
dVf 1 each individual specified by position and velocity vectors
( KE .VR  Vf ) (10)
dt TE (xi(t), vi(t)) at each time or iteration. Each individual is
dRf 1 named as a “particle” and the position of every particle
( KF .Vf  Rf ) (11) represents a potential solution to the under study
dt TF
optimization problem [10-13].
F. PID Controller Model
The basic algorithm of PSO is given by these 7 steps:
Each generator is generated by PID controllers. The PID
parameters are tuning with different methods, the classical 1. Create a population of agents (called particles) uniformly
Ziegler-Nichols method, Genetics algorithm (GA), Bacterial distributed over X.
Foraging optimization algorithm (BFOA), and with the 2. Evaluate each particle’s position according to the
optimization method (PSO). The diagram of the used PID objective function.
controller is shown in Fig.7 [10]. 3. If a particle’s current position is better than its previous
The PID model is given by: best position, update it.
K 4. Determine the best particle (the particle’s previous best
K (S ) K P  I  K D S (12)
S positions).
5. Update particles velocities according to:
Vit 1 Vit  C1rand1§¨ Pbest  X it ·¸  C2rand2 §¨ gbest  X it ·¸ (16)
© ¹ © ¹
6. Move particles to their new positions according to:
X it 1 X it  Vi t 1 (17)
7. Go to step 2 until stopping criteria are satisfied.
Fig.7. Block diagram of PID controller model.

The general algorithm of PSO is shown in Fig.8 [12-19]:

978-1-4799-8318-6/15/$31.00 ©2015 IEEE 426


3
Start
2

Frequency deviation (HZ)


1

Specify the parameters


0
of PSO
-1

Without AGC
-2
Generate initial population With AVR
With LFC
-3
With LFC and AVR (AGC)

-4
Time-Domain Simulation 0 10 20 30 40 50 60
Time (s)
Fig.9. AGC effect on the Frequency.
1
Find the fitness of each 0.9
particle in the current 0.8
population

Terminal Voltage (pu)


0.7

0.6

0.5
Gen= Gen+1
Stop 0.4
If Gen > Genmax Without AGC
0.3 With AVR
0.2 With LFC
With LFC and AVR (AGC)
0.1
Update the particle 0
0 10 20 30 40 50 60
position and velocity Time (s)
Fig.10. AGC effect on the Terminal Voltage.
Fig.8. PSO Algorithm. 5
Tie-line pow er flow deviation (MW)

0
IV. SIMULATION RESULTS
-5
This section presents the results of the proposed particle Without Controller
swarm optimization (PSO) approach by a comparison with -10 With PID Controller (Z-N)
With PID Controller (GA)
the results of genetics algorithm (GA), bacterial foraging -15 With PID Controller (PSO)
optimization (BFO) and the Ziegler-Nichols method [19]. With PID Controller (BFO)
A typical example of two-area power system is -20

considered for the simulation. The two-area power system is -25


simulated for 0.1pu step load perturbation in area-1 and the
values of the different parameters of the system are given in -30

Appendix. -35
0 5 10 15 20 25 30
The coupling effects between the AVR and LFC control Time (s)
loops as an AGC control system without PID controller are Fig.11. Tie-line power flow deviation.
shown in Fig.9 and Fig.10. The fluctuations in the tie-line
0.1
power flow, the system frequency and the terminal voltage
Frequency deviation in A rea-1 (H Z)

magnitude in each area using PID controller are shown in 0


Figs. 11-15 respectively.
-0.1

The PID controller parameters are given in Table I. -0.2

-0.3
TABLE I. PID CONTROLLER PARAMETERS.
Parameters -0.4
Kp Ki Kd Without Controller
-0.5 With PID Controller (Z-N)
Methods
With PID Controller (GA)
Ziegler-Nichols 0.4713 2.63 0.6575
-0.6 With PID Controller (PSO)
GA 0.9971 0.9775 0.8729
With PID Controller (BFO)
BFO 2.8718 1.7219 1.9680
-0.7
PSO 3.1205 2.7821 2.4641 0 5 10 15 20 25 30
Time (s)
Fig.12. Frequency deviation in Area-1.

978-1-4799-8318-6/15/$31.00 ©2015 IEEE 427


0.1
TABLE II. RESULTS AND COMPARISON.
Tuning PID controller 0.1 pu step load disturbance in area-1
Frequency deviation in Area-2 (HZ)

0
techniques Ziegler- GA BFOA PSO
-0.1
Nichols
-0.2 Max 0.1803 0.156 0.09355 0.04547
Frequency deviation[HZ]
-0.3
Area-1 Settling time 21.14 17.43 16.57 6.792
-0.4
Without Controller [s]
With PID Controller (Z-N)
-0.5
With PID Controller (GA)
Max 0.1254 0.1533 0.0868 0.007218
With PID Controller (PSO) Frequency deviation[HZ]
-0.6
With PID Controller (BFO) Area-2 Settling time 30 16.18 15.67 6.734
-0.7
0 5 10 15 20 25 30 [s]
Time (s) 5.52 5.297 2.548 0.4133
Max
Fig.13. Frequency deviation in Area-2. deviation[MW]
Tie-line
1.4 Settling time 30 22.13 20.43 2.036
[s]
Terminal Voltage in Area-1 (pu)

1.2
Terminal 17.82 17.82 17.82 17.82
Settling time
1 Voltage in
[s]
Area-1
0.8
Terminal Settling time 17.93 17.93 17.93 17.93
0.6 Voltage in
Without Controller [s]
With PID Controller (Z-N) Area-2
0.4
With PID Controller (GA)
With PID Controller (PSO)
0.2
With PID Controller (BFO) 0,2

Comparison [Hz]
0
0 5 10 15 20 25 30
Time (s) 0,15 ZieglerͲNichols
Fig.14. Terminal Voltage in Area-1.
1.2
0,1 GA
Terminal Voltage in Area-2 (pu)

1
0,05 BFOA
0.8
0 PSO
0.6
MaxFrequency
0.4 Without Controller DeviationinAreaͲ1
With PID Controller (Z-N)
0.2 With PID Controller (GA)
With PID Controller (PSO) (a). Max Frequency deviation in Area-1 [HZ].
0 With PID Controller (BFO)

-0.2
0 5 10 15 20 25 30
0,2
Time (s)
Comparison [Hz]

Fig.15. Terminal Voltage in Area-2.


0,15
The quality of power supply is determined by constant ZieglerͲNichols
of frequency and voltage. In this study, the combined 0,1 GA
proposed model between LFC and AVR named automatic
generation control (AGC) is applied for two-area 0,05 BFOA
interconnected power system. The real power and frequency
PSO
are adjusted by LFC loop, where as the reactive power and 0
voltage are regulated by AVR loop. The LFC is used to MaxFrequency
maintain a zero steady state error, while the AVR loop is to DeviationinAreaͲ2
maintain the machine output voltage within a specified
time. The system frequency and the tie-line power flow are
(b). Max Frequency deviation in Area-2 [HZ].
suppressed if both areas adopt LFC, and the fluctuations of
the voltage magnitude are eliminated and the voltage 6
magnitude is keeping at his reference value (Vref =1pu) if
Comparison [Mw]

5
the generators are equipped with AVR. The PID controller
is used to minimize the frequency over shoot and transient 4 ZieglerͲNichols
oscillations, also with optimal PID parameters the 3 GA
fluctuations of the system are suppressed most effectively. 2
Using the PSO method, the time of suppressing the BFOA
fluctuation (settling time) is very short compared with the 1
time given by the traditional Ziegler-Nichols method, 0 PSO
genetics algorithm (GA), and bacterial foraging MaxTieͲLinepower
optimization (BFO). The results are compared and the FlowDeviation
proposed approach (PSO) is proven to be better as shown in
Table II and Fig.16. (c). Max Tie-Line power flow deviation [Mw].

978-1-4799-8318-6/15/$31.00 ©2015 IEEE 428


35 TABLE.A.II AVR PARAMETERS.
Frequency TieͲLine
inAreaͲ2 PowerFlow AVR Parameters in Area-1 and Area-2
30
Frequency KA TA KE TE KF TF
SettlingTime[S]

25 inAreaͲ1 ZieglerͲ 10 0.1 1 0.4 2 0.04


20 Nichols
GA TABLE.A.III PSO PARAMETERS.
15 Population size 40
BFOA
10 Cognitive coefficient (C1) 1.6
PSO Social coefficient (C2) 0.4
5
Crossover rate (CR) 0.7
0
Maximum iteration 100
(d). Settling Time [s].
Fig.16. Comparative Analysis. REFERENCES
The convergence characteristic of PSO is depicted in [1] H. Bevrani, Robust Power System Frequency Control, Springer,
Fig.17, where the horizontal axis represents number of PSO 2009.
generation, and the vertical axis represents the minimum [2] H. Bevrani, T. Hiyama, Intelligent automatic generation control;
CRC press, Taylor & Francis 2011.
values of the calculated error ITAE presented in Eq. (15). [3] H. Saadat. Power System Analysis, McGraw-Hill, New York 1999.
1.4 [4] P. S. R. Murty, Operation and Control in Power Systems, BS
Publications 2008, ch.6-7.
1.2
[5] P. Kundur, Power System Stability and Control, McGraw-Hill, 1994,
1
pp. 581–626.
[6] A. Soundarrajan, Dr.S.Sumathi, C.Sundar, “Particle Swarm
Min ITAE

0.8 Optimization Based LFC and AVR of Autonomous Power


Generating System”, IAENG International Journal of Computer
0.6
Science, vol .37(1), 2010.
0.4 [7] M. Rahmani and N. Sadati, “Two-level optimal load–frequency
control for multi-area power systems”, Electrical Power and Energy
0.2 Systems, vol .53, 2013, pp .540–547.
[8] LI BiHui et al, “Study on the frequency control method and AGC
0
0 20 40 60 80 100 120 model of wind power integration based on the full dynamic process
Number of PSO Generation
simulation program”, IEEE International Conference on Advanced
Fig.17. Convergence Characteristic of PSO. Power System Automation and Protection, 2011, pp .246–251.
[9] J. G. Ziegler, N. B .Nichols, “Optimum settings for automatic
V. CONCLUSION controllers”, Trans ASME.Vol.64, 1942, pp.759-768.
[10] K. J. Astrom, T. Hagglund, “The future of PID control”, Control
In this paper a load frequency control (LFC) and Engineering Practice 2001, Vol.9, pp.1163–1175.
automatic voltage regulator (AVR) of two-area [11] Z-L. Gaing, “A particle Swarm Optimization approach for optimum
interconnected power system are studied. The coupling design of PID controller in AVR system”, IEEE Transactions on
Energy Conversion, Vol.19, No.2, 2004, pp384-391.
effects of the LFC and AVR control loops on the [12] J. Kennedy, R. Eberhart, “Particle Swarm Optimization”, Proceeding
fluctuations caused by step load disturbance are examined. of the IEEE International Conference on Neural Networks, Vol.IV,
An application of the particle swarm optimization (PSO) 1995, pp.1942-1948.
algorithm for determining the optimal values of the [13] Dr. K. RamaSudha, V.S. Vakula, R.Vijaya Shanthi, “PSO based
design of robust controller for two area load frequency control with
proportional-integral-derivation (PID) controller is applied. nonlinearities”, International Journal of Engineering Science and
The typical two-area system has been simulated for step Technology, Vol. 2, 2010, pp.1311-1324.
load disturbance in area-1. The results are compared to the [14] P.M Anderson, A.A Fouad, Power System Control and Stability, The
classical Ziegler-Nichols method, Genetics Algorithm (GA) Iowa State University Press, 1977.
and to the Bacterial Foraging Optimization (BFO); thereby [15] S. Satyanarayana, R.K. Sharma, Mukta, “Mutual Effect between LFC
and AVR Loops in Power Plant”, Electrical and Electronics
the proposed approach has proven to be very efficient. Engineering: An International Journal 2014, Vol .3(1), pp.61-69.
[16] A. J. Wood and B. F. Wollenberg, Power Generation Operation and
APPENDIX Control, 2nd ed., John Wiley & Sons, New York, 1966, pp. 328–362.
Data of a typical two-area power system. [17] Gurdeepinder Singh, Rajni Bala, “Automatic Generation & Voltage
Control of Interconnected Thermal Power System Including Load
Reference Frequency: f0= 50Hz; Inertia constant: M=10; Scheduling Strategy”, International Journal of Engineering and
Load-damping constant: D= 0.9 pu. Advanced Technology (IJEAT), Vol. 1, Issue 2, 2011.
[18] S. Panda, B. Mohanty, P. K. Hota, “Hybrid BFOA-PSO algorithm for
TABLE.A.I GOVERNOR-TURBINE PARAMETERS. automatic generation control of linear and nonlinear interconnected
power systems”, Applied Soft Computing 2013, Vol.13, pp. 4718–
Governor-Turbine Governor-Turbine
4730.
Area-1 Area-2
[19] N.EL.Y. Kouba, M. Menaa, M.Hasni, B. Boussahoua, M. Boudour,
KG1 20 KG2 16
“Optimal load frequency control based on hybrid bacterial foraging
Tsr1 0.2 Tsr2 0.3
and particle swarm optimization”, IEEE, 11th International Multi-
Tsc1 0.5 Tsc2 0.6 Conference on Systems, Signals & Devices, (SSD-PES), 2014,pp.1-6.

978-1-4799-8318-6/15/$31.00 ©2015 IEEE 429

You might also like