Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

International Conference on Recent Innovations in Electrical, Electronics & Communication Engineering - (ICRIEECE)

Design and analysis of 3-DOF PID controller for


load frequency control of multi area interconnected
power systems
Pritish Panda Aditya Kumar Nanda
M.Tech EE Student M.Tech EE Student
VSSUT, Burla VSSUT, Burla
India India
pritishpnd0@gmail.com aditya.nanda007@gmail.com

Abstract—In the present work, two-area three-source power peak loads. Due to change in load, errors in designing of
system is considered consisting of thermal, hydro and gas as the power system there is various unreliability in the power
generating units in the power system. The 3 degree of free- system. This prevents the capability of conventional control
dom proportional integral derivative controller (3-DOF PID) is
considered and gains of parameters are tuned by using a new strategies to control this unstable part of load frequency control
nature based grey wolf optimization (GWO) technique. (ITAE) is problem. Because of this traditional control strategies are not
taken as objective function. To show the ascendancy of proposed capable to handle this fluctuating part of the LFC problem. So,
approach, the results are compared with DE optimized PID and power systems must be designed to manage non-linearity and
TLBO optimized two degree of freedom (2-DOF) PID controller. inconsistent behavior so as to have a control over the real time
From the simulation, it was found that the results obtained are
better in terms of error, settling time and peak/under shoots. scenario. In the year of 1970, modern optimal control concept
Index Terms—3-Degree Of Freedom Proportional Integral for automatic generation control (AGC) in the power system
Derivative (3-DOF PID) Controller; Multi-source power system; was given by Fosha and Elgerd [4]. In this paper, critical
Multi-area power system; Load frequency control (LFC); Grey examination was done for Megawatt power and frequency
Wolf Optimization (GWO) control problem. Kumar P ,Kothari DP [5] gave a literature
survey on the AGC of power systems. Majhi S Parmer KPS
I. I NTRODUCTION [6] single area three sources power system comprising of
When there is some imbalance between generation of true reheating turbine thermal power plant, hydro power plant along
power and loading in the power system, it leads to variation with mechanical hydraulic governor and gas turbine power
in frequency of the system and power in the tie-line. This plant is considered and the proposed optimal output feed-
inequality between generation of power and demand is solved back controller provides satisfactory performance in reducing
by proper control of true power which is done by controlling frequency overshoot and transient oscillations than that of
frequency as true power depends on frequency of the system feedback controller. A generalized neural network approach
[1]. The prime role of Load Frequency Control is to keep for LFC in the power system has been proposed by Kalra PK,
desired real power output of a generating unit and help in Chaturbedi DK [7]. A new approach called as an approach
regulating the frequency of interconnected power system. It of reinforcement learning for AGC has been proposed by
is normally called as a constituent of AGC. Maintenance of Sastry PS, Rao PSN [8]. Mohanty et al. [9] considered a three
stability of power system can be done by keeping balance source power system consisting of thermal, gas and hydro
between real power and frequency. The net variation in gen- power plants and optimized the parameters of controller using
eration in each area is computed by LFC by monitoring the DE algorithm and found that DE optimized PID controller
frequency of the system and flow in the tie-line is referred provides better damping of oscillations in the system than
as Area Control Error (ACE).Low value of ACE is preferred that of optimal feedback controller. Algorithm for Differential
value which can be maintained by varying the set position of Evolution was developed by Stron Price [10]. The research was
the alternators present in the particular area. When the ACE is further extended by Sahu Rabindra Kumar et. al [11] and a
made zero by LFC, then tie-line power as well as frequency new controller was introduced called as 2-Degree Of freedom
errors will lead to zeros [2]. Control areas or regions consist (DOF) PID controller and optimized the controller parameters
of set of generators in a vast power system. The integration using Teaching Learning Based Optimization (TLBO) algo-
of hydro, thermal, nuclear, gas, renewable energy resources rithm and found that this controller produces better response
etc. results in a control area [3]. Thermal generation and as compared to PID controller optimized by DE algorithm for
hydro generation is normally used to meet the base loads load frequency control (LFC) in three source two area power
and gas power generation is kept to meet peak demands or system. Algorithm for TLBO was given by Rao et.al [12].
978-1-5386-5995-3/18/$31.00 ©2018 IEEE

2983
International Conference on Recent Innovations in Electrical, Electronics & Communication Engineering - (ICRIEECE)

II. D ESIGNING OF THE SYSTEM UNDER STUDY discharge volume time constant of turbine are represented as
XC , YC , bg, cg, TCR , TF , TCD respectively.
Multi source two area interconnected power system is given
in Fig. 1.In this dynamic model, each area is provided with
III. T HE 3 DOF PID CONTROLLER
three inputs consisting of controller input (∆Pref ), disturbance
in load (∆PD ) and error in the tie-line power (∆PT ie ) and two The independent adjustment of number of closed loop
outputs consisting of deviation in frequency (∆F ) and Area transfer functions is known as degree of freedom (DOF).The
control error (ACE).This ACE is equated in Eq.(1). internal structure of 2-DOF PID controller is shown in fig. 2.
ACE = B∆F + ∆PT ie (1)

In the given model, parameters for regulating speed of the

Fig. 2. Internal Structure of two DOF Controller

Fig. 1. Transfer Function model of two area three source power system Fig. 3. Internal Structure of three DOF Controller

governor for thermal , hydro , gas generating plants are The fig. 3 presents the internal design of 3-DOF PID controller.
represented as R1 , R2 , R3 (p.u. Hz) respectively; similarly pa- As it is a three degree of freedom so, it includes three control
rameters of frequency bias are represented as B1 , B2 , B3 . Area loops in the system. Here, R(s) represents input which is the
control error is represented by ACE1 , ACE2 , and ACE3 . In ACE of respective area. Y(s) indicates the frequency deviating
thermal power plant, the time constant of speed governor, the in each area and output of the controller is given as C(s)
time constant of turbine, the turbine reheat constant, steam which act as input to the generating units. The basic difference
turbine reheat time constant are represented as TSG , TT , Kr between 2-DOF and 3-DOF is the disturbance D(s) in the
and Tr respectively. In hydro power plant, speed governor system which is given as feedback in 3-DOF and it is absent
main servo time constant of turbine, nominal starting time in 2-DOF PID controller. This additional loop present in
of water in penstock, speed governor transient droop time the controller helps in eliminating disturbance in the system
constant of turbine ,speed governor reset time of turbine are quickly which further improves the stability in the system.
represented as TGH , TW , TRH and TRS respectively. In gas The objective function is decided as per the requirements of
turbine power plant , lead time constant of speed governor of the system and limitations in the system. The main aim of
turbine ,lag time constant of speed governor of turbine, turbine objective function is to have best values for the parameters of
constant ,valve position of turbine , combustion reaction time the controller. Here, objective function used is ITAE because
delay of turbine , fuel time constant of turbine , compressor of the following advantages depicted below.

2984
International Conference on Recent Innovations in Electrical, Electronics & Communication Engineering - (ICRIEECE)

A. Benefits of ITAE the members of this group is inducted to alpha if any


1) The settling time is much quicker than Integral Squared of the alpha grows old or dies. Beta pack of wolves
Error (ISE). always have regards for the alpha group and lead the
2) It drastically reduces the overshoot in the system. lower groups in the hierarchy.
3) Delta Group When a wolf doesn’t come under alpha or
So, because of the above reasons ITAE is taken as objective
beta or omega then it is classified as Delta. These pack
function. Eq.(2) depicts the expression for ITAE.
of wolves consist of old wolves, guardian of the group
Z tsim
and efficient hunters. They alert the group in case of risk.
IT AE = (| ∆F1 | + | ∆F2 | + | ∆Ptie |).t.dt (2) They work under the guidance of alpha and beta but lead
0
the omega.
where,
4) Omega Group The omegas are present at lowest in the
∆F1 = Variation of frequency in area 1
rank of hierarchy. They have to obey the orders of alpha,
∆F2 = Variation of frequency in area 2
beta and omega group. They are responsible for taking
∆Ptie = Variation in Tie-line power
care of the baby wolves present in the group.
tsim = Simulation time
A. Mathematical Modelling of Grey Wolf Optimization
The objective function needs to be minimized using
i Social Ranking As per the ranking of wolves, alpha
following constraints.
group of wolves are considered as best solutions followed
KPmin ≤ KP ≤KPmax
by beta, which is followed by delta and the remaining
KImin ≤KI ≤KImax
solutions are considered as omega.
KDmin ≤KD ≤KDmax
ii Surrounding the Prey Grey wolves surround the prey
PWmin ≤ PW≤ PWmax
while going to attack the prey. Surrounding of prey can
DWmin ≤ DW≤DWmax
be mathematically expressed as:
Nmin ≤ N≤ Nmax
~ =| C.
B ~ Y~p (t) − Y
~ (t) | (3)
KPmin and KPmax = Minimum and maximum gain of
proportional controller respectively. ~ (t + 1) = Y~p (t) − A.
Y ~B ~ (4)
KImin and KImax = Minimum and maximum gain of integral
controller respectively. ~ = 2~ak~1 − ~a
A (5)
KDmin and KDmax = Minimum and maximum gain of
~ = 2.k~2
C (6)
derivative controller respectively
PWmin and PWmax = Minimum and maximum proportional where
weight respectively. t = presesnt iteration
DWmin and DWmax =Minimum and maximum derivative A~ and C~ = Coefficient Vectors
weight respectively ~
Yp (t) = Vector position of Prey
Nmin and Nmax =Derivative filter coefficient minimum and ~ (t) = Vector Position of Grey Wolf
Y
maximum limits.
~a = Value decreased from 2 to 0
k1 and k2 = Random values in the range of 0 to 1
IV. G REY W OLF O PTIMIZATION T ECHNIQUE
Using eqn (4) and (5) , upgradation of position inside the
Grey Wolf Optimization (GWO) Algorithm was first coined search space around the prey is done by the grey wolf in
by Mirjalili [13].Grey Wolves are classified under families of any random location .
Canidae. Grey wolves in the apex of food chain. These wolves iii Hunting of Prey Alpha group of wolves take care of
are always in pack. The size of the group is between six to hunting process and sometimes Beta and Delta participate
twelve. They are further classified as in it and this three gives best solutions as per their ranking.
1) Alpha Group This can be mathematically depicted as:
2) Beta Group ~1 = Y~α − A ~ 1 .B

Y (7)
3) Delta Group
4) Omega Group ~ α =| C
B ~ 1 .Y
~α − Y
~ | (8)
1) Alpha Group leaders of the pack are considered as
~2 = Y
Y ~β − A
~ 1 .B
~β (9)
Alphas that play a vital role in taking decisions, place of
hunting, place of sleeping. The pack of wolves follows ~ β =| C
~ 1 .Y
~β − Y
~ |
B (10)
the decision taken by Alphas.
2) Beta Group The next group of wolves are classified as ~3 = Y
Y ~δ − A
~ 1 .B
~δ (11)
beta group. These group of wolves aid alphas in taking
decisions and other activities related to the pack. One of ~ δ =| C
B ~ 1 .Y
~δ − Y
~ | (12)

2985
International Conference on Recent Innovations in Electrical, Electronics & Communication Engineering - (ICRIEECE)

~ ~ ~
Y~ (t + 1) = Y1 + Y2 + Y3 (13)
3
Eqn. (7), (9) and (11) gives best position of alpha wolves,
beta wolves and delta wolves respectively and the final position
is found by taking the average of position of the these three
wolves which is given by eqn. (13).

V. R ESULTS AND D ISCUSSION


The newly designed 3DOF-PID controller is implemented
in two area three source power system and the obtained results
are compared with pre-published results [9] and [11] which is
shown in Fig.4, Fig 5 and Fig 6 respectively.

Fig. 6. Change in tie-line power

chart format representation of settling time, peak overshoot,


peak undershoot of the controllers are shown in Fig.7, Fig.8,
Fig.9 respectively.

Fig. 4. Frequency response in area-1

Fig. 7. Chart format representation of Settling time

Fig. 5. Frequency response in area-2 Fig. 8. Chart format representation of Overshoot

When GWO optimized 3DOF-PID controller is used, it was When 3DOF PID controller optimized by Grey Wolf Opti-
found that it gives much improved results as compared to pre- mization was used in two area three source power systems, it
published results in terms of settling time, peak overshoot and was found that it gives much more improved results as com-
peak undershoot which is shown in table 1.The ITAE values pared to pre-published results such as Differential Evolution
of the various optimized controller is given in table 2.The Optimized PID controller and Teaching Learning based 2DOF
parameters of the designed controller is given in table 3.The PID controller. This can be seen through superior results in

2986
International Conference on Recent Innovations in Electrical, Electronics & Communication Engineering - (ICRIEECE)

2.4Hz/p.u.; TT = 0.3s, TSG = 0.08s, Kr = 0.3, Tr =


10s, TP S = 1.49s; T12 = 0.0433, a12 = 1, , KT =
0.543478; KH = 0.326084; KG = 0.130438.

TABLE III
ITAE VALUES FOR VARIOUS OPTIMIZED CONTROLLER

Sl.No Type of controller ITAE


1 DE optimized PID controller 0.4474
2 TLBO optimized 2DOF PID controller 0.2343
3 GWO optimized 3DOF PID controller 0.1744

Fig. 9. Chart format representation of Undershoot


R EFERENCES
TABLE I [1] Elgerd OI. Electric energy systems theory. An introduction. 2nd ed. New
O PTIMIZED VALUE OF 3DOF C ONTROLLER Delhi: Tata McGraw-Hill; 2007
[2] Kundur P. Power system stability and control; 8th reprint. New Delhi:
Parameters GWO based 3DOF PID Tata McGraw-Hill; 2009.
KP 1 -0.3237105 [3] Hassan B. Robust power system frequency control. New York: Springer;
KI1 0.001456855 2009.
KD1 0.3156752 [4] Elgerd OI, Fosha CE. Optimal megawatt-frequency control of multi-area
PW1 4.416055 electrical energy systems. IEEE Trans PAS 1970;89(4):556–63.
DW1 0.6203145 [5] Ibraheem, Kumar P, Kothari DP. Recent philosophies of automatic
N1 280.3849 generation control strategies in power systems. IEEE Trans Power Syst
D1 0.01 2005;20:346–57
KP 2 0.01481201 [6] Parmar KPS, Majhi S, Kothari DP. Load frequency control of a realistic
KI2 -0.0007554611 power system with multi-source power generation. Int J Elect Power
KD2 1.708486 Energy Syst 2012; 42:426–33
PW2 0.0148243 [7] Chaturvedi DK, Satsangi PS, Kalra PK. Load frequency control:
DW2 2.110205 a generalize neural network approach. Electr Power Energy Syst
N2 97.63254 1999;21:405–15.
D2 0.01 [8] Ahamed TPI, Rao PSN, Sastry PS. A reinforcement learning approach
to automatic generation control. Electr Power Syst Res 2002; 63:9–26.
[9] Panda S, Mohanty B, Hota PK. Controller parameters tuning of dif-
ferential evolution algorithm and its application to load frequency
terms of settling time, peak overshoot and peak under shoot. control of multisource power system. Int J Electr Power Energy Syst
Also, it gives better ITAE values than that of earlier two 2014;54:77–85.
results. So, it can be concluded that the proposed controller [10] Stron R, Price K. Differential evolution – a simple and efficient adaptive
scheme for global optimization over continuous spaces. J Global Optim
is more efficient than that of the controllers with which the 1995;11:341–59.
results are compared. [11] Kumar, Sahu Rabindra and Panda, Sidhartha and Rout, Umesh and
Kumar Sahoo, Dillip (2016). Teaching learning based optimization
algorithm for automatic generation control of power system using 2-DOF
TABLE II PID controller. International Journal of Electrical Power and Energy
P ERFORMANCE OF THE S YSTEM Systems. 77. 287-301.
[12] RV, Savsani VJ, Vakharia DP. Teaching–learning-based optimization: an
DE based TLBO based GWO based optimization method for continuous non-linear large scale problems. Inf
Variations Parameters
PID 2-dof PID 3-dof PID Sci 2012;183(1):1–15.
Settling time 12.3404 10.562 6.741 [13] Syed Mirjalili et al, “Grey Wolf Optimizer” Advances in Engineering
Area 1 Peak Software 69 (2014) 46–61
-0.0284 -0.00853 -0.00156
Undershoot
Peak
0.0026 0.000576 0.000255
Overshoot
Settling time 8.2911 11.4377 5.756
Area 2 Peak
-0.024 -0.000869 -0.000639
Undershoot
Peak
0.0068 0.0054 0.00354
Overshoot
Settling time 14.0627 10.0456 5.658
Tie-Line Peak
-0.0046 -0.0028 -0.00021
Undershoot
Peak
0.00023 0.00021 0.00015
Overshoot

A PPENDIX
R1 = R2 = R3 = 2.4Hz/p.u.; TGH = 0.2s, TW =
1.0s, TRH = 28.75s, bg = 0.05s, cg = 1, TF = 0.23s, YC =
1s, XC = 0.6s, TCD = 0.2s, TCR = 0.01s, TRS =
5s, KP S = 68.9566Hz/p.u.M W ; R1 = R2 = R3 =

2987

You might also like