You are on page 1of 88

Manuscript Details

Manuscript number IJEPES_2018_2415_R2

Title Dual loop IMC structure for load frequency control issue of multi-area multi-
sources power systems

Article type Research paper

Abstract
In this paper, a dual loop-internal model control (DL-IMC) scheme for load frequency control (LFC) issue of multi-area
power systems is presented. The proposed scheme contains two control loops for disturbance rejection and
minimization of oscillations. The inner loop of the proposed configuration contains an IMC controller and the outer loop
controller is derived by using a predictive model based on model reduction. In order to validate the proposed
approach, single and two area reheated thermal power system (TPS), and hydrothermal power systems (HTPS) are
considered. The robustness of the proposed scheme is shown by introducing perturbations in the system parameters
for different values of external load disturbances. The proposed method is extended for a three-area reheated HTPS
with wind and solar power penetrations along with random wind speed and solar irradiance. In addition to this, the
simulation results of a four-area reheated HTPS with renewable power sources such as wind power generator, solar
power generator, fuel cell and aqua-electrolyzer to show the effectiveness of the proposed approach.

Keywords internal model control; load frequency control; model reduction; multi-area power
systems.

Corresponding Author Deepak Kumar

Corresponding Author's Motilal Nehru National Institute of Technology Allahabad, India


Institution

Order of Authors Bheem Sonker, Deepak Kumar, Paulson Samuel

Suggested reviewers Victor Sreeram, Jayanta Pal, S. P. Singh, Surya Prakash

Submission Files Included in this PDF


File Name [File Type]
cover letter.docx [Cover Letter]

Response_new_08032019.docx [Response to Reviewers]

Paper_revision_15-3-19_marked.docx [Revised Manuscript with Changes Marked]

Highlights.docx [Highlights]

Paper_revision_15-3-19_not marked.docx [Manuscript File]

To view all the submission files, including those not included in the PDF, click on the manuscript title on your EVISE
Homepage, then click 'Download zip file'.

Research Data Related to this Submission


There are no linked research data sets for this submission. The following reason is given:
No data was used for the research described in the article
15-03-2019

To,

The Editor-in-chief
International Journal of Electrical Power & Energy Systems

Dear Sir,

Please find our revised manuscript entitled “Dual loop IMC structure for load frequency
control issue of multi-area multi-sources power systems” authored by Bheem Sonker, Deepak
Kumar, and Paulson Samuel for possible publication in the International Journal of Electrical
Power & Energy Systems.
We have attached “Responses to reviewers’ comments’ file also. We are extremely thankful
to you and the reviewers for kind evaluation of the manuscript and the constructive
suggestions to improve quality of the paper. We have tried our best to incorporate the
suggested changes in the revised manuscript. The incorporated corrections in the revised
submitted manuscript of 1st revision are indicated in BLUE colour and the current revision is
shown in GREEN colour.
Thank you very much.

Yours sincerely

Deepak Kumar
Review Response

We are extremely thankful to honourable editor and the reviewers for their kind evaluation of
the manuscript and the constructive suggestions to improve quality of the paper. We have
tried our best to incorporate the suggested changes in the revised manuscript. The
incorporated corrections in the revised submitted manuscript of 1st revision are indicated in
BLUE colour and the current revision is shown in GREEN colour.

Comments of Reviewer 1 and responses:


The author has made an attempt to develop an Dual loop IMC structure for load frequency
control issue of multi-area power systems. It is welcome, I appreciate the idea. In addition, it
seems that the manuscript is weak on the current state of-the-art, because author has taken
Old two area power system.
Response: Thank you very much for the comments. In the 1st revision of the manuscript, we
had shown simulation results of proposed approach for the following models:
(i) Single area reheated thermal power system (TPS),
(ii) Two-area reheated TPS
(iii) Single area reheated hydrothermal power systems (HTPS)
(iv) Two area reheated HTPS
(v) Four-area reheated HTPS with renewable power sources such as wind power generator,
solar power generator, fuel cell and aqua-electrolyzer

In addition to this, in the current revised version of the manuscript, we have added the
simulation results for a three-area reheated HTPS with wind and solar power penetrations
along with random wind speed and solar irradiance.

Comment 1: The paper is more an academic exercise than a practical contribution.


Response: As per the comment, in order to improve the impact of the manuscript, new
simulation results of a three-area reheated hydrothermal power systems with wind and solar
power penetrations along with random wind speed and solar irradiance, have been added. It
is observed from the simulation results of all the considered power system models (listed as
above) that the proposed method performs well for removal of disturbance and frequency
deviations.
Comment 2: The main drawback of the paper is author fails to explain the need of Dual
IMC in AGC to control frequency and power deviations. Author should explain briefly
the need of IMC as compare with the existing controllers.
Response: The proposed control structure is a dual feedback loop configuration for load
frequency control. The inner loop controller is designed by a disturbance-rejection controller
(DRC) QD ( s ) in the inner loop (IL), a predictive model Pm ( s ) in series with QD ( s ) in the

forward path, and an outer loop controller (OLC) GC ( s ). The OLC is incorporated in the

system for removing external load disturbances which gives an uncontrolled system output.
The predictive model is designed using model reduction method which is further used to
design QD (s ) and GC ( s ). Therefore, the proposed approach rejects very efficiently than the

existing methods as shown in the simulation results of Section 4. The above discussion has
been given in the first paragraph of Section 3.

Comment 3: The main motivation is justified through simulations, result and


comparison.
Response: Thank you very much for the comment.

Comment 4: The author has considered old primitive two area model to test the
efficacy of the controller, but it is not sufficient to publish in this standard journal.
Response: Thank you very much for the comment. In previous version of the manuscript, we
had shown simulation results of proposed scheme for the following models:
(i) Single area reheated thermal power system (TPS),
(ii) Two-area reheated TPS
(iii) Single area reheated hydrothermal power systems (HTPS)
(iv) Two area reheated HTPS
(v) Four-area reheated HTPS with renewable power sources such as wind power
generator, solar power generator, fuel cell and aqua-electrolyzer

The current version of the manuscript also includes the simulation results of a three-area
reheated HTPS with wind and solar power penetrations along with random wind speed and
solar irradiance.
Comment 5. To modify the paper author should consider Wind and solar energy
penetration, Random wind speed, Random irradiance change, also should be extended
to three area model.
Response: Thank you very much for the comment. As per the suggestion, the proposed work
has been extended for a three-area reheated hydrothermal power systems with wind and
solar power penetrations along with random wind speed and solar irradiance. It is observed
from the simulation results that the proposed method reduces the frequency and tie-line
power deviations with fewer oscillations and gives a satisfactory load disturbance rejection
as compared to the without controller case.

Comment 6. The paper has to be well organised.


Response: As per the suggestion, in order to organize the manuscript, we have shifted the

block diagram of the two area reheated thermal power system (TPS) into Section 2 as Fig. 2.

Further, we have arranged the simulation results in the following order:

Section 4.1 Single area reheated TPS,

Section 4.2 Two-area reheated TPS

Section 4.3 Single area reheated hydrothermal power systems (HTPS)

Section 4.4 Two area reheated HTPS

Section 4.5 Three-area reheated HTPS extension with wind and solar energy penetrations

Section 4.6 Four-area reheated HTPS extension with renewable power generations

Comment 7. There are also other works in AGC with optimised controller has been
addressed by different authors. So the paper should have sufficient novelty to publish in
this journal.
Response: Thank you very much for the comment. As per the best of our knowledge, the
proposed control scheme is completely novel and it has not been developed/ implemented by
other researchers till date. In the originally submitted manuscript, the proposed scheme was
validated for
(i) Single area reheated thermal power system (TPS),
(ii) Single area reheated hydrothermal power systems (HTPS)
(iii) Two area reheated HTPS
(iv) Two-area reheated TPS

The simulation results were included by introducing variations of 50% in the parameters of
the power system models to validate the robustness of the proposed method. All results are
reproduced for nominal values, lower bounds and upper bounds of the system parameter
values.

In response to the comments of first round review, the results of Two-area reheated TPS
extension were compared with a recently proposed Bacterial Foraging-optimized fuzzy
PI/PID controller for AGC method [R1]. Further, the results of a four-area reheated HTPS
with renewable power sources such as wind power generator, solar power generator, fuel
cell and aqua-electrolyzer.
In the current revision, we have also added the results of a three-area reheated HTPS with
wind and solar power penetrations along with random wind speed and solar irradiance.

[R1] Arya Y, Kumar N. Design and analysis of BFOA-optimized fuzzy PI/PID controller for
AGC of multi-area traditional/restructured electrical power systems. Soft Computing 2017;
21: 6435–6452.

Comment 8: After clear examination of simulation and results it can be concluded


that results are appreciable but it is not sufficient, the work must be extended to three
area with wind, solar penetrations.
Response: Thank you very much for the comment. As per the suggestion, the proposed work
has been extended for a three-area reheated hydrothermal power systems with wind and
solar power penetrations along with random wind speed and solar irradiance in newly
included Section 4.5. It is observed from the simulation results that the proposed method
reduces the frequency and tie-line power deviations with fewer oscillations and gives a
satisfactory load disturbance rejection as compared to the without controller case.

Comments of Reviewer 2:
The authors have addressed all of the proposed comments, so I propose to be accepted
for its publication.
Response: Thank you very much.

Comments of Reviewer 3:
No more comments.
Response: Thank you very much.
Dual loop IMC structure for load frequency control issue of multi-
area multi-sources power systems
Bheem Sonker, Deepak Kumar, and Paulson Samuel
Department of Electrical Engineering, MNNIT Allahabad, Prayagraj, UP, India
*Correspondence: deepak_kumar@mnnit.ac.in

Abstract: In this paper, a dual loop-internal model control (DL-IMC) scheme for load frequency

control (LFC) issue of multi-area power systems is presented. The proposed scheme contains two

control loops for disturbance rejection and minimization of oscillations. The inner loop of the

proposed configuration contains an IMC controller and the outer loop controller is derived by

using a predictive model based on model reduction. In order to validate the proposed approach,

single and two area reheated thermal power system (TPS), and hydrothermal power systems

(HTPS) are considered. The robustness of the proposed scheme is shown by introducing 50%

perturbations in the system parameters for different values of external load disturbances. The

proposed method is extended for a three-area reheated HTPS with wind and solar power

penetrations along with random wind speed and solar irradiance. In addition to this, the

simulation results of a four-area reheated HTPS with renewable power sources such as wind

power generator, solar power generator, fuel cell and aqua-electrolyzer to show the effectiveness

of the proposed approach.

Keywords: internal model control; load frequency control; model reduction; multi-area power

systems.

Appendix A. Nomenclature

f , f1 , f 2 , Incremental frequency deviations TH 1 , TH 2 , Hydro turbine time


in Hz TH 3 , TH 4 ,
constants (s)
f3 , f 4
Pd , Pd 1, Pd 2 , External load disturbance in p.u. u1 ( s ), u2 ( s ) Control outputs of
MW controllers
Pd 3, Pd 4 ,
K p , K p1 , K p 2 , Electric system gains G p , G p1 , G p 2 , Transfer function of load
Hz/p.u.MW Gp3 , Gp 4 and machines
K p3 , K p 4
Tp , Tp 1 , Tp 2 Electric system time constants (s) Gg , Gg 1 , Gg 2 , Transfer function of steam
Gg 3 , Gg 4 governor
Tp 3 , Tp 4
Tt , Tt 1 , Tt 2 , Turbine time constants (s) Gr , Gr1 , Gr 2 , Transfer function of steam
Tt 3 , Tt 4 Gr 3 , Gr 4 reservoir
Tg , Tg 1 , Tg 2 , Governor time constants (s) Gt , Gt1 , Gt 2 , Transfer function of steam
Gt 3 , Gt 4 turbine
Tg 3 , Tg 4 ,
R, R1 , R2 , Speed regulations due to GHg1 , GHg 2 , Transfer function of hydro
R3, R4 , R5, governor action in Hz/p.u. MW GHg 3 , GHg 4 governor

R6 , R7, R8 ,
Tr , Tr1 , Tr 2 , Reheated time constants (s) GHR1 , GHR 2 , Transfer function of hydro
Tr 3 , Tr 4 GHR 3 , GHR 4 , reservoir

K r , K r1 , K r 2 , High pressure turbine power GHw1 , GHw 2 , Transfer function of hydro


Kr3 , Kr 4 fraction GHw3 , GHw 4 turbine

Tw1 , Tw 2 , Water time constants (s) PTie Tie-line power error


Tw3 , Tw 4
THg 1 , THg 2 , Hydro governor time constants (s) T12 , T23 , Synchronizing coefficients
THg 3 , THg 4 T34 , T41
THR1 , THR 2 , Hydro reservoir time constants (s) Pr1 , Pr 2 , Power system rated area
THR 3 , THR 4 Pr 3 , Pr 4 capacity (MW)
K wpg1 , K wpg 2 , Wind power generator (WPG) K spg1 , K spg 2 , Solar power generator
gains (SPG) gains
K wpg 3 , K wpg 4 K spg 3 , K spg 4
K fc1 , K fc 2 , Fuel cell (FC) gains K ae1 , K ae 2 , Aqua- electrolyzer (AE)
K ae3 , K ae 4 gains
K fc 3 , K fc 4
Twpg1 , Twpg 2 , WPG time constants (s) Tspg1 , Tspg 2 , SPG time constants (s)
Twpg 3 , Twpg 4 Tspg 3 , Tspg 4
T fc1 , T fc 2 , FC time constants (s) Tae1 , Tae 2 , AE time constants (s)
T fc 3 , T fc 4 Tae 3 , Tae 4

1. Introduction
During operation of a large-scale power system, the fluctuations caused by load demand

may affect the system frequency and tie-line power (TLP) interchange considerably. Therefore, a

load frequency controller is required to maintain the frequency and TLP exchange at the

scheduled values. In general, a load frequency controller maintains the performance of a power

system during external load disturbance (ELD) and parametric uncertainties [1]. Since the last

few decades, several approaches using different control strategies have been proposed for LFC of

single area and multi-area power systems (MAPS) during normal operating conditions and load

perturbations. The IMC is one of the popular approaches for LFC of single area and MAPS.

Using two degrees of freedom-internal model control (TDF-IMC) scheme, Tan [2-3] designed a

proportional-integral-derivative (PID) controller for LFC. Further, PID controller based

problems and solutions for LFC are presented in [4]. Later, Saxena and Hote [5] presented a

TDF-IMC based approach for LFC of single area non-reheated TPS. Various structures of TDF-

IMC for eliminating load disturbance and its applications can be referred from [6]. A PID

controller is designed for LFC of MAPS in [7] and a modified IMC is presented in [8] for a

better performance of the closed-loop system. Xiao et al. [9] presented an IMC based PID

controller for power control in the frequency domain. Another IMC based PID controller is

introduced by Ghousiya et al. [10] using an optimal H2 minimization. A modified TDF-IMC

scheme for LFC of a non-reheated TPS using model reduction (MR) is presented in [11].

Recently, Singh et al. [12] presented a TDF-IMC based PID controller for LFC of the single area

and multi-area reheated HTPS via logarithmic approximations. The PID controllers for LFC

have also been designed by direct synthesis approach using frequency response matching,

imperialist competitive algorithm and Laurent series expansion in [13], [14] and [15]. In addition

to this, Sathya et al. [16] designed PI controllers with dual mode gain scheduling and Dash et al.
[17] presented PD-PID controller using BAT inspired algorithm for an interconnected power

system. Rahman et al. [18] introduced an optimized 3DOF-PID controller using the

biogeography-optimized technique for LFC of a TPS under a deregulated environment. Later,

Rahman et al. [19] presented another approach for automatic generation control (AGC) of

unequal four-area TPS. In [20], a PID plus second order derivative controller for AGC of MAPS

is designed using the ant-lion optimizer algorithm. Subsequently, Sahu et al. [21-22] designed

TDF-PID controllers using optimization algorithm based on teaching-learning for AGC of a

power system and designed tilt integral derivative controller with a filter for LFC of MAPS.

Furthermore, LFC using fractional order-PID controllers are presented in [23], [24] and a

distributed model predictive control based LFC for MAPS using discrete-time Laguerre

functions is presented in [25]. The observer-based sliding mode control approaches for load

frequency control are presented in [26] and [27]. An artificial neural network and fuzzy-based

adaptive neuro-fuzzy inference system (ANFIS) approach for AGC of three area HTPS is

presented in [28]. Some other fuzzy based approaches for load frequency control are presented

using a modified whale optimization in [29], bacterial foraging optimization in [30], optimal

controller in [31], ultra-capacitor and thyristor control phase shifter in [32] and a neuro-fuzzy

intelligent controller in [33].

It is well known that a multi-area multi-source power system results into a higher order

mathematical model which poses several issues in designing of a suitable load frequency

controller. Such controllers give a sluggish response and require more computational effort.

Therefore, it is desired to use a reduced order mathematical model to design controller such that

improved performance may be obtained. Although several controllers using IMC have been

presented for LFC problem, there is a large scope for improving the performance with
disturbance rejection and load changes issue. Therefore, this article presents the DL-IMC scheme

for LFC of MAPS as an extension of recent work [34] by employing a MR method [35]. The

contributions of the paper are as follows:

1) A reduced order model of the original power system model is obtained which is used as a
predictive model in the proposed approach.
2) A controller is developed by incorporating pole clustering and improved Padѐ

approximation [35] to the proposed scheme for achieving less computational effort.

3) The robustness of the proposed approach is verified by introducing 50% variations in

the parameters of the power system model.

4) Performance indices such as integral square error (ISE), integral absolute error (IAE) and

integral time absolute error (ITAE) are computed with original and perturbed system

parameters for a comparative analysis of the proposed approach.

To show the effectiveness of the proposed method, it is compared with existing methods [3,

13, 15, 23, 30] for single area reheated TPS and with [12, 30] for single area and two area

HTPS. It is observed that the proposed scheme minimizes the frequency deviations and TLP

exchange better than the existing techniques. Further, the proposed scheme also maintains the

robustness under random load variations which is shown by introducing different values of

ELD as well as parametric uncertainties. The article is arranged as follows: a brief

description of single area reheated TPS, and single and two area reheated HTPS are included

in Section 2 whereas the proposed scheme is explained in Section 3. The simulation results of

the proposed scheme with nominal and random load changes are discussed in Section 4.

Finally, the conclusion is presented in Section 5.

2. Power system models under consideration


In this section, single area and multi-area multi-source power system models have been

considered for validation of the proposed approach. A single area TPS [1, 3, 13, 15, 23] with a

reheated turbine is considered as a first system model as shown in Fig. 1. The mathematical

model of a single area reheated TPS consists of linearized models of the steam governor, steam

reservoir, reheated turbine, and load and machines. A reheated turbine provides higher thermal

efficiency and the transfer function of a reheated turbine involves an additional time constant Tr1

along with the turbine time constant. The second considered model is a two area TPS [30] which

has reheated turbine as shown in Fig. 2. The third model is a single area HTPS [12] which has

multi-source generation with reheated and hydro turbines as shown in Fig. 3. The mathematical

model of a single area reheated TPS consists of linearized models of the steam governor, steam

reservoir, reheated turbine, hydro governor, hydro reservoir, hydro turbine along with load and

machines. The fourth model is a two area multi-unit multi-source HTPS [12] with reheated and

hydro turbines in each area of the model as shown in Fig. 4.

Pd
_
u+ 1 (1  sK r1Tr1 ) 1 +
Kp f
_ (1  sTg ) (1  sTr1 ) (1  sTt ) (1  sTp )
Gg Gr Gt Gp
1
R

Fig. 1. Single area reheated TPS


1
B1 R1

Pd 1
+ ACE1 _ _
+ 1 (1  sK r1Tr1 ) 1 + K p1 f1
Controller
+ (1  sTg1 ) (1  sTr1 ) (1  sTt1 ) _ (1  sT p1 )
Gg1 Gr1 Gt1 G p1

+
T12
s _

+ ACE 2 _
+ 1 (1  sK r 2Tr 2 ) 1 + K p2 f 2
Controlller
_ _ (1  sTg 2 ) (1  sTr 2 ) (1  sTt 2 ) _ (1  sT p 2 )
Gr 2 Gg 2 Gt 2 Gp2

1 Pd 2
B2
R2

Fig. 2 Two area reheated TPS

1 (1  sK r1Tr1 ) 1 Pd 1
(1  sTg 1 ) (1  sTr1 ) (1  sTt1 ) _
u1 + Gg 1 Gr1 Gt1 + K p1 f1
_ + (1  sT p1 )
+
1 (1  sTHR1 ) (1  sTw1 )
G p1
_ (1  sTHg 1 ) (1  sTH 1 ) (1  0.5sTw1 )
1 GHg 1 GHR1 GHw1
R1 1
R2

Fig. 3. Single area reheated HTPS


1
R1 1
B1
R2
1 (1  sK r1Tr1 ) 1 Pd 1
+ ACE1 _ (1  sTg 1 ) (1  sTr1 ) (1  sTt1 ) _
Controller
+ Gg 1 Gr1 Gt1 + K p1 f1
+ _ + _ (1  sT p1 )
+
1 (1  sTHR1 ) (1  sTw1 )
G p1
(1  sTHg 1 ) (1  sTH 1 ) (1  0.5sTw1 )
GHg 1 GHR1 GHw1
+
T12
s _
_
_

1 (1  sK r 2Tr 2 ) 1
+ ACE 2 (1  sTg 2 ) (1  sTr 2 ) (1  sTt 2 ) _
+ Gg 2 Gr 2 Gt 2 + K p2 f 2
Controlller
_ _ + _ (1  sT p 2 )
+
1 (1  sTHR 2 ) (1  sTw 2 )
Gp2
_ (1  sTHg 2 ) (1  sTH 2 ) (1  0.5sTw 2 )
1 GHg 2 GHR 2 GHw 2 Pd 2
B2 1
R3
R4

Fig. 4. Two area reheated HTPS

The transfer functions (TFs) of each block (refer Appendix A for the values of parameters) are

shown in the figures. The overall TF of reheated TPS G ( s ) (refer Fig. 1) with respect to load

disturbance as reference input is given by

f Gp  s
G  s   (1)
Pd 1  G s G s G ( s )G s 1
p  g   r t  
R

It is clear that the LFC is a load disturbance rejection problem which uses the feedback law

u( s )  Gc ( s )f ( s ) to stabilize the power plant G ( s ) under load disturbance Pd ( s ) and

reduces the effect of Pd ( s).

The TF of area 1 for the two area reheated TPS (refer Fig. 2), is obtained with respect to

external load disturbance as reference input which is given by


f1 G p1 ( s ) B1
G2 rtps ( s )   (2)
Pd 1 1  G ( s )G ( s )G ( s )G ( s ) 1
p1 g1 r1 t1
R1

The TF of single area reheated HTPS (refer Fig. 3) is derived with respect to external load

disturbance as reference input which is as below

f1 G p1 ( s )
G1 ( s )   . (3)
Pd 1 1  G p1 ( s )((Gg1 ( s )Gr1 ( s )Gt1 ( s ) / R1 )  (GHg1 ( s )GHR1 ( s )GHw1 ( s ) / R2 ))

For the two-area case, the area control errors and TLP expressions are written as

e1 (t )  ACE1  B1f1  PTie ,


e2 (t )  ACE2  B2 f 2  PTie , (4)
T12
PTie  (f1  f 2 ).
s
where B1 and B2 are the frequency bias constants (p.u.MW/Hz), ACE1 and ACE2 are the area

control errors of area 1 and area 2 respectively. The area controls errors are used to regulate the

output power of the generator as per the ELD. From Eq. (3), it is clear that the LFC is the main

issue against RLD and variation in ELD. The feedback control law for area 1 and 2 are obtained

as follows

u1 ( s )  GC1 ( s ) ACE1 ,
(5)
u2 ( s )  GC 2 ( s ) ACE2 ,

where GC1 ( s ) and GC 2 ( s ) are the feedback controllers. In two area case, a decentralized

controller can be obtained by assuming that there is no TLP exchange i.e. PTie  0. Based on

the above equations, the TF of area 1 for the two area reheated HTPS (refer Fig. 4), is obtained

with respect to external load disturbance as reference input as

f1 G p1 ( s ) B1
G2 ( s )   , (6)
Pd 1 1  G p1 ( s )((Gg1 ( s )Gr1 ( s )Gt1 ( s ) / R1 )  (GHg1 ( s )GHR1 ( s )GHw1 ( s ) / R2 ))

Thus, the controllers for LFC of each area can be obtained similarly.
3. Proposed scheme

The proposed DL-IMC scheme is designed to reduce the effects of ELD and parametric

uncertainties during operation of a MAPS. The control block diagram of the proposed scheme is

shown in Fig. 5(a) which consists of a plant model G ( s ), a disturbance-rejection controller

(DRC) QD ( s ) in the inner loop (IL), the PM Pm ( s ) in series with QD ( s ) in the forward path, and

an outer loop controller (OLC) GC ( s ) in negative feedback path. In order to simplify the control

block diagram, the summing junction before QD ( s ) can be separated in two parts. The first part

gives a cascade connection of Pm ( s ) and QD ( s ) resulting into the forward path controller (FPC)

K m ( s )  Pm ( s )QD ( s ) as shown in Fig 5(b) that is a simplified version of Fig. 5(a). The second

part consists of unity gain in forward path and QD ( s ) in feedback path which gives

K d ( s )  1 / (1  QD ( s )) as the series controller (SC). The OLC is incorporated in the system for

removing ELD which causes the system output to become uncontrolled. In order to stabilize the

uncontrolled response of the system caused by OLC, an FPC is incorporated in the proposed

approach. However, the FPC produces oscillations in the system, which is minimized by the

addition of an IL controller that also helps in fast removal of ELDs. Further, a suitable model

reduction method is necessary to obtain the predictive model Pm (s ) which is used to design

QD (s) and GC ( s ). The tuning parameter  d used for design of QD (s) must be chosen properly. As

the design of OLC is based on the TDF-IMC scheme, it gives faster rejection of external load

disturbances and the inner loop controller reduces the overall gain of the closed loop system.
Pd (s )
_ Pd (s )
R (s ) + + C (s ) _
G(s )
_ _ R (s ) + + + C (s )
G(s ) K d (s )
_ _
Pm (s ) QD (s )
K m (s )
+ +

GC (s ) GC (s )

Fig. 5. a) Proposed DL-IMC scheme b) Simplified version of proposed scheme

The PM and conventional IMC configurations are used for developing the DL-IMC

scheme and the PM is derived by using an MR technique [35] which is discussed as below:

3.1 Derivation of the PM

A generalized k th order TF G ( s ) for single area reheated TPS, single area HTPS and each

of two areas HTPS is considered as

g 0  g1s  .....  s k 1 g k 1
G (s)  . (7)
h0  h1s  h2 s 2  ....  s k hk

The PM Pm ( s )  N ( s ) D( s ), of the proposed scheme is derived from the system model G ( s ) by

using pole clustering and improved Padѐ approximation [35] for the denominator and numerator

of PM respectively. The steps to obtain D( s ) are as follows:

1. Sort the real poles i.e. p1  p2  p3 ......  pv and complex poles ( 1  j1 ,  2  j 2 …..

 v 1  j v 1 ) i.e. 1   2  .......   v 1 and 1   2  .......   v 1 of G ( s ).

2. Set   1;

3. Compute the pole cluster centers (PLC) of real poles and complex poles as

1 1 1
 v  1    v 1  1    v 1  1  
 j       v  , Aej       (v  1)  and Bej       (v  1)  . (8)
 i 0  pi    i 0  i    i 0   i  
4. Set     1;

5. Generate another PLC as

1 1 1
  1 1     1 1     1 1  
       2  , Ae      2  and Be    *   2  . (9)
 p1  1  
   1 L 1  
   1 L 1  
 

6. If   v  1 for real pole and   v for complex poles, then go to step-7. Otherwise, go to

step 4.

7. Finally, the cluster centers are obtained as pe1  v 1 for real pole, e*1  Aev  jBev and

e1  Aev  jBev for complex poles. Therefore, the denominator of PM is obtained as

D ( s )  ( s  pe1 )( s  e*1 )( s  e1 )  x0  x1s  x2 s 2  x3s 3 (10)

Further, N ( s ) is determined by matching time moments and Markov parameters of G ( s ) and

PM is obtained by expanding G ( s ) as below:


G ( s )   M i s  i 1 (about s  )
i 0
(11)

   Ti s i
(about s  0),
i 0

where Ti and M i are the i th Time moment and Markov parameters of G ( s ) respectively. If

N ( s )  n0  n1s  n2 s 2 , then

g 0  h0T0 ,
g1  h0T1  h1T0 ,
n0  x0T0 , (12)
n1  x0T1  x1T0 ,
n2  x3 M 0 .

Therefore, the PM is obtained as

N ( s) n0  n1s  n2 s 2
Pm ( s )   . (13)
D ( s ) x0  x1s  x2 s 2  x3s 3
3.2 Design of controllers

The controllers of the proposed DL-IMC scheme are designed by using the conventional

IMC configuration [3, 5] which is shown as below:

1. Factorize Pm ( s ) as

Pm ( s )  Pm  ( s ) Pm  ( s ), (14)

where Pm  ( s ) and Pm  ( s ) are the minimum-phase and non-minimum-phase parts.

2. Design the IMC controller QD ( s ) as

QD ( s )  Pm1 ( s )Qd ( s ), (15)

1  1s   2 s 2   3s 3
where Qd ( s )  is the third order low pass filter,  d is the tuning
( d s  1) n 1

parameter and n is the total number of poles Pm ( s ) that are p1 , p2 ,  , pn . The values of

1,  2, and  3 are obtained by

1  1s   2 s 2   3s 3
(1  ) |s  p1, p2, .... pn  0. (16)
( d s  1) n 1

3. The design of OLC GC ( s ) is based on the complementary sensitivity transfer function

T ( s ) of the IMC feedback configuration [8] which is given by

T ( s )  Pm ( s )QD ( s )
 Pm ( s ) Pm1 ( s )Qd ( s )
Pm  ( s )(  3s 3   2 s 2  1s  1) (17)
 .
( d s  1) n 1
P ( s )GC ( s )
As T ( s )  m , we get GC ( s ) from (15) as
1  Pm ( s )GC ( s )

Pm1 ( s )(  3s 3   2 s 2  1s  1)
GC ( s )  . (18)
{( d s  1) n 1  Pm  ( s )(  3s 3   2 s 2  1s  1)}
4. Simulation Results

The performance and robustness of the proposed method for LFC of single area reheated

TPS, single and two areas reheated HTPS are discussed in this section by varying ELD.

4.1. Single-area reheated TPS

A single-area reheated TPS [3] is considered as shown in Fig. 1 with the following system

parameters:

K p  120, Tp  20, Tt  0.3, Tg  0.08, R  2.4, Tr  4.2, and K r  0.35.

After substituting the values in Eq. (1), the TF of system G ( s ) is obtained as

6s 3  96.42 s 2  272.62 s  59.53


G( s)  (19)
s 4  16.12 s 3  46.42 s 2  48.65s  25.3

The poles of system are 12.79, 2, 0.665  0.7395i. As discussed in Section 3.2, the controllers

QD ( s ) and GC1 ( s ) are designed by using minimum phase Pm  ( s ) of the PM Pm ( s ) where the

Pm ( s ) is obtained by using Section 3.1 as

N ( s) 6 s 2  23.147 s  5.21
Pm ( s )   3 (20)
D ( s ) s  3.566 s 2  3.963s  2.211

The step responses of G ( s ) and Pm ( s ) PM are compared as shown in Fig. 6. It is clear from

the figure that the response of predictive model completely follows the path of G ( s ) and it also

achieves approximately zero steady-state error. Therefore, this predictive model can be used for

designing the controllers.

Remark 1: The model reduction method must be chosen such that the resulting reduced model

matches the steady state response of the original model. Further, the reduced model must be

stable.
Step Response
6

Amplitude
3

System model
1
Predictive model
0
0 2 4 6 8 10 12
Time (seconds)

Fig. 6. Comparison of step responses between system model and predictive model

After substituting, the values  d  0.05 and n  3 in Eq. (15) and Eq. (18) respectively, the

DRC QD ( s ) is obtained as

( s 3  3.566 s 2  3.963s  2.211)(0.0005s 3  0.015s 2  0.2 s  1)


QD ( s )  , (21)
(6 s 2  23.147 s  5.21)(0.05s  1) 4

where  3  0.0005,  2  0.015 and 1  0.2 .

Finally, OLC GC ( s ) is obtained as

0.0005s 6  0.01678s 5  0.2555s 4  1.774 s 3  4.392 s 2  4.405s  2.211


GC ( s )  . (22)
0.000037 s 6  0.0001447 s 5  0.00003256s 4  8.087e  7 s 3  1.808e 17 s 2

An ELD Pd  0.01 is applied at t  2 sec. to the system for determining performance of

the proposed method. Further, the robustness of the proposed method is analysed by inserting

50% variations in the parameters of system. Hence, the parameters of single area reheated TPS

are expressed as

K p  [60,180], Tp  [10,30], Tt  [0.15, 0.45], Tg  [0.04, 0.12], Tr  [2.1, 6.3],


K r  [0.175, 0.525] and R  [1.2,3.6];

The responses of the proposed scheme are shown in Figs. 7(a)-7(c) for nominal value

(NV), lower bound (LB) and upper bound (UB) of the system parameters. The proposed method
is compared with different controllers such as fractional order PID [23], IMC-PID [3], PID [13,

15]. It is clear from the figures that the proposed method removes ELDs with fewer oscillations

and settles much quicker than the existing techniques [3, 13, 15, 23]. The performance of the

proposed scheme and existing techniques are also compared through the error indices for

nominal and perturbed parameters. The expressions for the error indices are given as follows:


(a) ISE   e(t ) dt.
2

0

(b) IAE   e(t ) dt.
0

(c) ITAE   t e(t ) dt.
0

It is observed from Tables 1, 2 and 3 that the performance indices of the proposed scheme are

significantly lower than the existing techniques [3, 13, 15, 23]. Therefore, the proposed DL-IMC

scheme effectively handles the frequency deviations and rejects the disturbances very well

during power system operation.

Remark 2: The tuning parameter  d required in the design of DRC needs to be chosen by taking

a suitable compromise between closed-loop response and robustness. A large value of tuning

parameter results into a sluggish closed-loop response and appropriate robustness towards the

variation in the system parameters. On the other hand, a smaller value generates a comparatively

faster closed-loop response and inappropriate robustness.

Remark 3: A suitable optimization algorithm can also be used to obtain the tuning parameters

used in the inner and outer loop controllers to improve the performance of the proposed

approach. However, it is avoided in the current work to simplify the proposed method.
Remark 4: It is to be noted that the proposed approach can also be used for other stable
processes and it works well. In the case of unstable processes, an additional compensator may be
required.

(a) (b)

(c)
Fig. 7. Frequency deviations (FD) of the proposed scheme for single area reheated TPS of 1% load variation with (a)

NV, (b) LB and (c) UB.

Table 1
Single area reheated TPS for nominal value

Error indices
Methods 1% Load variation
ISE IAE ITAE
Proposed method 1.436×10-8 5.962×10-5 4.678×10-4
Tan’s method [3] 6.152×10-5 1.118×10-2 7.419×10-2
Anwar & Pan’s [13] 6.938×10-6 3.992×10-3 2.672×10-2
Padhan & Majhi’s [15] 2.002×10-5 5.722×10-3 3.972×10-2
Sondhi & Hote’s [23] 1.594×10-5 3.937×10-3 2.883×10-2

Table 2
Single area reheated TPS for LB

Error indices
Methods 1% Load variation
ISE IAE ITAE
Proposed method 1.377×10-8 5.873×10-5 4.606×10-4
Tan’s method [3] 4.823×10-5 7.828×10-3 5.597×10-2
Anwar & Pan’s [13] 5.466×10-6 3.636×10-3 2.284×10-2
Padhan & Majhi’s [15] 1.595×10-5 4.959×10-3 3.334×10-2
Sondhi & Hote’s [23] 1.476×10-5 4.199×10-3 2.938×10-2

Table 3
Single area reheated TPS for UB

Error indices
Methods 1% Load variation
ISE IAE ITAE
Proposed method 1.449×10-8 6.005×10-5 4.71×10-4
Tan’s method [3] 5.746×10-5 1.05810-3 7.043×10-2
Anwar & Pan’s [13] 7.572×10-6 4.244×10-3 2.683×10-2
Padhan & Majhi’s [15] 2.356×10-5 6.231×10-3 4.2×10-2
Sondhi & Hote’s [23] 1.618×10-5 3.74×10-3 2.74×10-2

4.2. Two-area reheated TPS extension

The proposed method is implemented for two-area reheated TPS power system [30] as

shown in Fig. 2. The values of parameters of system are given as below:

K p1  K P 2  120, Tp1  Tp 2  20, Tt1  Tt 2  0.3, Tg1  Tg 2  0.08, Tr1  Tr 2  10, K r1 


K r 2  0.5, R1  R2  2.4, B1  B2  0.425, T12  T21  0.545, Pr1  Pr 2  2000.

After substituting the values in Eq. (2), we obtain the TF models of area 1 and area 2 of two

areas reheated TPS. The transfer function G2 rtps ( s ) of area 1 of two area reheated TPS with

respect to disturbance input is obtained as


2.55s 3  40.625s 2  110.3s  10.625
G2 rtps ( s )  (23)
s 4  15.94 s 3  40.042 s 2  58.42 s  10.625

and the predictive model is obtained by using Section 3.1 as

N (s) 2.55s 2  4.903s  0.4702


Pm 2 rtps ( s )   3 (24)
D( s ) s  3.023s 2  2.607 s  0.4702

The proposed controllers for LFC of two-area reheated TPS power system are obtained as

( s 3  3.023s 2  2.607 s  0.4702)(0.0002 s 3  0.0092 s 2  0.16 s  1)


QD ( s )  and
(2.55s 2  4.903s  0.4702)(0.04 s  1) 4

0.0002 s 6  0.009805s 5  0.1883s 4  1.508s 3  3.444 s 2  2.682 s  0.4702


GC ( s )  .
0.000006528s 6  0.0001554 s 5  0.001296 s 4  0.001988s 3  0.0001881s 2

A 1% ELD of full load is applied at t  2 sec. to area 1 of the system and the performance of

proposed method is analysed by introducing 50% parametric uncertainties as

K p1  K p 2  [60,180], Tp1  Tp 2  [10,30], Tt1  Tt 2  [0.15, 0.45], Tg1  Tg 2  [0.04, 0.12],


Tr1  Tr 2  [5,15], K r1  K r 2  [0.25, 0.75], and R1  R2  [1.2,3.6].

The simulation results are shown in Figs.8 (a)-8(c) for NV, LB and UB and it is observed from
the figures that the proposed scheme provides significantly better results than the recently
proposed fuzzy PI controller [30].

(a) (b)
(c)
Fig. 8. (a) FD of area 1, with 1% load variations in area 1for NV, LB and UB (b) FD of area 2, with 1% load
variations in area1 for NV, LB and UB, (c) TLP exchange with 1% load variations in area 1 for NV, LB and UB.

4.3. Single-area reheated HTPS

The values of parameters for single-area reheated HTPS (refer Fig. 3) are considered from

[12] as

K p1  120, Tp1  20, Tt1  0.3, Tg1  0.08, Tr1  10, K r1  0.5, Tw1  1, THg1  41.6, THR1  5,
TH 1  0.513, R1  R2  2.4.

After substituting the values in Eq. (3), the TF model G1 ( s ) obtained as

6s 6  119.4 s 5  663.3s 4  1438s 3  1145s 2  124.2 s  2.343


G1 ( s )  7 . (25)
s  19.96s 6  109.2 s 5  262.1s 4  352.6s 3  364.8s 2  66.83s  1.972

As discussed in Section 3.2, QD1 ( s ) and GC1 ( s ) are determined by using Pm1 ( s ) that is obtained

from Pm1 ( s ) . Hence, from the approach given in Section 3.1, the 3rd order PM Pm1 ( s ) is obtained

as

N ( s) 0.05116  2.3065s  6s 2
Pm1 ( s )   . (26)
D ( s ) 0.04306  1.118s  0.8763s 2  s 3
A comparison of step responses of single area reheated HTPS G1 ( s ), PM Pm1 ( s ) , and the

reduced models by Singh et al. [12] is shown in Fig. 9 and it is observed that the utilized MR

approach produces a satisfactory PM.

Step Response
6
G1(s): System model
5 Pm1(s): Predictive model
R1 [12]: Singh et al.
R2 [12]: Singh et al.
4
Amplitude

0
0 5 10 15 20
Time (seconds)
Fig. 9. Comparison of step responses between the original system and reduced models

After substituting, the values  d  0.1 and n  3 in Eq. (15) and Eq. (16) respectively, the

DRC QD1 ( s ) is obtained as

( s 3  0.8763s 2  1.118s  0.04306)(0.0039 s 3  0.0599 s 2  0.4 s  1)


QD 1 ( s )  , (27)
(6 s 2  2.3065s  0.05116)(0.1s  1) 4
where  3  0.0039,  2  0.0599 and 1  0.4 .

Finally, OLC GC 1 ( s ) is obtained as

0.0039 s 6  0.06332 s 5  0.4569 s 4  1.418s 3  1.326s 2  1.135s  0.04306


GC1 ( s )  . (28)
0.0006s 6  0.0008307 s 5  0.0008358s 4  0.0002358s 3  5.116e 06 s 2

A 1% load variation in full load i.e. 2000MW at t  2.0 sec. is applied to the system and

the robustness of the proposed method is analysed by inserting 50% parametric uncertainties as

K p  [60,180], Tp  [10,30], Tt1  [0.15, 0.45], Tg1  [0.04, 0.12], Tr1  [5,15],
K r1  [0.25, 0.75], Tw1  [0.5,1.5], THg1  [20.8, 62.4], THR1  [2.5, 7.5], TH 1  [0.2565, 0.7695]
and R1  [1.2,3.6].
The corresponding results are shown in Figs. 10(a)-10 (c) for NV, LB and UB. It is

observed from the results that ELD rejection is very fast and smoother as compared to Singh et

al.’s technique [12].

(a) (b)

(c)
Fig. 10. Frequency deviations (FD) of the proposed scheme for single area reheated HTPS with (a) NV (b) LB and

(c) UB.

The performance indices of the proposed method are expressed in Tables 4, and it is clear

that the proposed method gives significantly lower values than the existing techniques [12].

Therefore, the proposed DL-IMC rejects the disturbances very fast and maintains the robustness.

Table 4
Single area reheated HTPS
Error 1% Load variation
indices NV LB UP NV LB UP
Singh et al. IMC-PID [12] Proposed method
ISE 1.604×10-5 1.948×10-5 1.838×10-5 2.823×10-8 2.664×10-8 2.914×10-8
IAE 3.939×10 -3 5.145×10 -3 3.969×10-3 1.029×10-4 1.007×10-4 1.05×10-4
ITAE 2.0×10-2 3.564×10-2 2.879×10-2 5.886×10-4 7.759×10-4 8.103×10-4

4.4. Two-area reheated HTPS

The values of parameters for two area reheated HTPS (refer Fig. 4) are taken from [12] as

K p1  K P 2  120, Tp1  Tp 2  20, Tt1  Tt 2  0.3, Tg1  Tg 2  0.08, Tr1  Tr 2  10, K r1 


K r 2  0.5, Tw1  Tw 2  1, THg1  THg 2  41.6, THR1  THR 2  5, TH 1  TH 2  0.513, R1  R2
 R3  R4  2.4, B1  B2  0.425, T12  T21  0.545, Pr1  Pr 2  2000.

After substituting the values in Eq. (6), we obtain the TF models of area 1 and area 2 of two

areas reheated HTPS. The analysis of only one area is shown as we know that both TFs come out

to be equal for the same values of parameters. The TF model G2 ( s ) of area 1 is obtained as

2.55s 6  50.74 s 5  281.9 s 4  611.1s 3  486.6s 2  52.8s  0.99


G2 ( s )  . (29)
s 7  19.96s 6  109.2 s 5  262.1s 4  352.6s 3  364.8s 2  66.83s  1.972

From Section 3.2, the PM Pm 2 ( s ) is obtained as

N ( s) 0.02162  0.9816 s  2.55s 2


Pm 2 ( s )   . (30)
D ( s ) 0.04306  1.118s  0.8763s 2  s 3

Similar to the previous case, after substituting, the values  d  0.2 and n  3 in Eq. (15)

and Eq. (17) respectively, the DRC QD 2 ( s ) is obtained as

( s 3  0.8763s 2  1.118s  0.04306)(0.0306s 3  0.2382 s 2  0.7999 s  1)


QD 2 ( s )  , (31)
(2.55s 2  0.9816s  0.02162)(0.2 s  1) 4
where  3  0.0306,  2  0.2382 and 1  0.7999.

Finally, OLC GC 2 ( s ) for two area case is obtained as


0.0306s 6  0.265s 5  1.043s 4  1.969 s 3  1.781s 2  1.152 s  0.04306
GC 2 ( s )  . (32)
0.00408s 6  0.005141s 5  0.005999 s 4  0.002052 s 3  0.0001371s 2  2.162e 06 s

To check the robustness of the proposed scheme, the parametric uncertainties are considered for

area 1 and area 2 and the responses of the proposed scheme are obtained as follows:

Case 1. The ELD of 2% is applied at t  2 sec. to area 1 of the system for NV and the

simulation results are shown in Figs. 11(a)-11(c). It is observed from the figures that the

proposed scheme provides the better results than the methods used in [12].

(a) (b)

(c)

Fig. 11. (a) FD of area-1 with 2% in area 1 for NV, (b) FD of area-2 with 2% load variation in area 1 for NV, and (c)

TLP exchange with 2% load variation in area 1 for NV.


Case 2. A 2% ELD of full load is applied at t  2 sec. to area 1 of the system and the robustness

of the proposed scheme is checked by inserting 50% parametric uncertainties as

K p1  K p 2  [60,180], Tp1  Tp 2  [10,30], Tt1  Tt 2  [0.15, 0.45], Tg1  Tg 2  [0.04, 0.12],


Tr1  Tr 2  [5,15], K r1  K r 2  [0.25, 0.75], Tw1  Tw 2  [0.5,1.5],
THg1  THg 2  [20.8, 62.4], THR1  THR 2  [2.5, 7.5], TH 1  TH 2  [0.2565, 0.7695]
and R1  R2  R3  R4  [1.2,3.6].

It is observed from the Figs. 12(a)-12(c) for LB and Figs. 13(a)-13(c) for UB that the proposed

scheme gives better disturbance rejection than [12].

(a) (b)

(c)
Fig. 12. (a) FD of area 1, with 2% load variations in area 1for LB, (b) FD of area 2, with 2% load variations in area1

for LB, (c) TLP exchange with 2% load variations in area 1 for LB.
(a) (b)

(c)

Fig. 13. (a) FD of area 1, with 2% load variations in area 1for UB, (b) FD of area 2, with 2% load variations in area1

for UB, (c) TLP exchange with 2% load variations in area 1 for UB.

The performance of the proposed scheme and existing techniques are compared through

the error indices for nominal parameters and parametric uncertainties of two-area reheated

HTPS. It is observed from Tables 5, 6 and 7 that the performance indices of the proposed scheme

are significantly lower than the existing techniques [12]. Therefore, the proposed scheme gives

better disturbance rejection and robustness. The tie-line powers of two area reheated HTPS by

using proposed and existing methods [12] are shown in Table 8 and it is clear that the proposed

method reduces the tie-line power exchange as compared to [12].


Case 3. The load change of amplitude 0.15, -0.2, 0.2, and 0.1 are applied at t= 5, 15, 25, and 35

sec. to area 1 for analyzing the robustness of the proposed scheme. It can be observed from Figs.

14(a)-14(d) that the proposed scheme provides better robustness than [12].

(a) (b)

(c) (d)
Fig. 14. (a) Load changes (b) FD of area 1 (c) FD of area 2 (d) TLP exchange of two-area reheated HTPS by load

changes at area 1.

Remark 5: The system parameters of the power system models can be obtained by parameter

identification methods presented in literature such as [36- 38]. However, the system parameters

used in this work are referred from [3, 12] for simplicity.

Table 5
Two areas reheated HTPS for NV

2% load variation in area 1


Methods Area 1 Area 2 Tie-line
( f1 ) ( f 2 ) ( Ptie )

ISE IAE ITAE ISE IAE ITAE ISE IAE ITAE


7.139 7.683 5.755 6.58 2.887 2.074 7.527 1.051 7.503
Proposed method
×10-7 ×10-4 ×10-3 ×10-8 ×10-4 ×10-3 ×10-9 ×10-4 ×10-4
Singh et al. IMC- 7.47 1.35 8.496 2.647 9.069 5.449 2.573 2.54 1.63
PID [12] ×10-5 ×10-2 ×10-2 ×10-5 ×10-3 ×10-2 ×10-6 ×10-3 ×10-2
Singh et al. 5.291 1.243 7.523 1.556 8.309 4.464 1.154 1.809 1.134
Conv. PID [12] ×10-5 ×10-2 ×10-1 ×10-5 ×10-3 ×10-2 ×10-6 ×10-3 ×10-2

Table 6
Two areas reheated HTPS for LB

2% load variation in area 1


Methods Area 1 ( f1 ) Area 2 ( f 2 ) Tie-line ( Ptie )

ISE IAE ITAE ISE IAE ITAE ISE IAE ITAE


7.086 7.655 5.734 5.519 2.541 1.842 5.996 8.813 6.381
Proposed method
×10-7 ×10-4 ×10-3 ×10-8 ×10-4 ×10-3 ×10-9 ×10-5 ×10-4
Singh et al. IMC- 1.039 1.749 1.067 4.395 1.254 7.232 3.804 3.1 2.012
PID [12] ×10-4 ×10-2 ×10-1 ×10-5 ×10-2 ×10-2 ×10-6 ×10-3 ×10-2
Singh et al. 7.503 1.722 9.593 2.871 1.205 6.095 1.868 2.367 1.472
Conv. PID [12] ×10-5 ×10-2 ×10-2 ×10-5 ×10-2 ×10-2 ×10-6 ×10-3 ×10-2

Table 7
Two areas reheated HTPS for UB

2% load variation in area 1


Methods Area 1 ( f1 ) Area 2 ( f 2 ) Tie-line ( Ptie )

ISE IAE ITAE ISE IAE ITAE ISE IAE ITAE


7.307 7.857 5.879 5.78 2.601 1.885 6.159 8.912 6.454
Proposed method
×10-7 ×10-4 ×10-3 ×10-8 ×10-4 ×10-3 ×10-9 ×10-5 ×10-4
Singh et al. IMC- 7.302 1.183 7.621 2.064 7.038 4.383 2.055 2.203 1.43
PID [12] ×10-5 ×10-2 ×10-2 ×10-5 ×10-3 ×10-2 ×10-6 ×10-3 ×10-2
Singh et al. 5.662 1.249 7.452 1.378 6.909 3.818 1.02 1.712 1.069
Conv. PID [12] ×10-5 ×10-2 ×10-2 ×10-5 ×10-3 ×10-2 ×10-6 ×10-3 ×10-2

Table 8
Tie-line power of two area reheated HTPS

TLP exchange (p.u. MW) (Ptie )


Methods 2% load variation in area 1
NV LB UB
Proposed method -1.244×10-8 -1.605×10-8 -7.615×10-9
Singh et al. IMC-PID [12] 2.54×10-5 5.038×10-6 1.002×10-5
Singh et al. IMC-PID [12] -4.483×10-6 -1.828×10-5 -1.38×10-5
4.5. Three-area reheated HTPS extension with wind and solar energy penetrations

The proposed method is extended to three-area multi-source power system, which consists of

reheated HTPS [12] with wind and solar energy penetrations by considering random wind speed

and random irradiance change [39, 40] as shown in Fig. 15(a)-15(d). The predictive model,

DRC, OLC and the values of parameters of three-area reheated HTPS, are referred from the two-

area reheated HTPS as given in Section 4.3. A wind power generator with random speed [40] is

considered as shown in Fig. 15(c). The mechanical torque of wind generator is developed by

wind energy which transfers through shaft to rotor of generator. The net output power of

generator is developed by random wind speed which is used to reheat the HTPS. The WPG

mathematical equations are referred from [39]. The solar power penetration model is designed as

shown in Fig. 15 (d). The fluctuation in solar power is obtained by using solar irradiance model

which uses a white noise block.

Reheated HTPS+Wind and Reheated HTPS+Wind and


Solar penetrations Solar penetrations

AREA-2 AREA-3

AERA-1

Reheated HTPS+Wind and


Solar penetrations

Fig. 15 (a) Simplified model of three-area power system


Pwi Ppvi
Wind power Solar power
+ +
Reheated thermal power

1 (1  sK riTri ) 1 Pdi
(1  sTgi ) (1  sTri ) (1  sTti ) _
ui + K pi
+ Ggi Gri Gti + fi
_ + _ (1  sTpi )
+
1 (1  sTHRi ) (1  sTwi )
G pi
_ (1  sTHgi ) (1  sTHi ) (1  0.5sTwi )
1 GHgi GHRi GHwi
Ri 1
Rj Hydro power

Bi
+
ACEi Tij
PTie s _

f j i
Fig. 15 (b) Block diagram of ith -area power system

+ 1
Wind speed
_ s

Saturation Saturation
_ + K Pitch angle Wind Torque
Constant
_ s turbine
+

+
Generator speed K1
s _
Saturation Generator model

_
+
 + K2 
_ s
Power

Lookup table Output power

Fig. 15 (c) Wind power penetration model


1
12 s  1
 +  K spg Solar power

+ Tspg s  1
White-noise Low-pass filter 

0.6*sqrt

Psolar initial

Fig. 15 (d) Solar power penetration model

The random load disturbance, random wind speed and solar irradiance are applied to the

model, which are plotted in Fig. 16(a)-16(c). The random load disturbance of the amplitudes 0.1,

-0.1, 0.15 and 0.05 are applied at t = 10, 30, 50, and 70 seconds to area 1 for analyzing the

robustness of the proposed scheme.

In Section 4.4, the proposed method is implemented for LFC of two area reheated HTPS

without any external renewable power sources and better results are achieved with fewer

oscillations. Therefore, in this section, the proposed method is applied for LFC of three-area

reheated HTPS with WPG and SPG power along with random wind speed and solar irradiance.

The simulation results of proposed method for three-area reheated HTPS along with random

wind speed and solar power penetration is shown during the random load disturbance for two

cases: (i) with proposed controller; and (ii) without proposed controller. It is observed from Figs.

16(a)-16(f) that the proposed method is capable of reducing the frequency and tie-line power

deviations with minimum oscillations significantly and gives a satisfactory load disturbance

rejection as compared to the without controller case.


(a)

(b) (c)
Fig. 15 Plots of (a) random load disturbance (b) wind speed and power, (c) Solar irradiance and power
(a) (b)

(c) (d)
(e) (f)
Fig. 16 Responses of FD and TLP of three-area multi-unit multi-sources reheated HTPS with WPG and SPG (a)
Area-1 (b) Area-2 (c) Area-3 (d) Area-4, and (e)- (h) Tie-line power for three area.

4.6. Four-area reheated HTPS extension with renewable power generations

The proposed method is extended to four-area multi-source power system, which consists of

reheated HTPS [12] with renewable power generations (RPGs) such as wind power generation

(WPG), solar power generation (SPG), fuel cell (FC) and aqua-electrolyzer (AE) as shown in

Fig. 17(a)-17(b). The models of renewable power sources are referred from [41-44] which is a

series combination of linearized models of WPG ( Gwpg ), SPG ( Gspg ), FC ( G fc ) and AE ( Gae ).

The values of parameters of system are given as below:


K p1  K P 2  K p 3  K P 4  120, Tp1  Tp 2  Tp 3  Tp 4  20, Tt1  Tt 2  Tt 3  Tt 4  0.3,
Tg1  Tg 2  Tg 3  Tg 4  0.08, Tr1  Tr 2  Tr 3  Tr 4  10, K r1  K r 2  K r 3  K r 4  0.5,
Tw1  Tw 2  Tw3  Tw 4  1, THg1  THg 2  THg 3  THg 4  41.6, THR1  THR 2  THR 3  THR 4  5,
TH 1  TH 2  TH 3  TH 4  0.513, R1  R2  R3  R4  R5  R6  R7  R8  2.4,
B1  B2  B3  B4  0.425, T12  T23  T34  T41  0.545, Pr1  Pr 2  Pr 3  Pr 3  2000.
K wpg1  K wpg 2  K wpg 3  K wpg 4  1, K spg1  K spg 2  K spg 3  K spg 4  1,
K fc1  K fc 2  K fc 3  K fc 4  1/100, K ae1  K ae 2  K ae3  K ae 4  1/ 500,
Twpg1  Twpg 2  Twpg 3  Twpg 4  1.5, Tspg1  Tspg 2  Tspg 3  Tspg 4  1.8,
T fc1  T fc 2  T fc 3  T fc 4  4, Tae1  Tae 2  Tae3  Tae 4  0.5;

Reheated HTPS+RPGS Reheated HTPS+RPGS

AREA-2 AREA-3

AERA-1 AREA-4

Reheated HTPS+RPGS Reheated HTPS+RPGS

Fig. 17 (a) Simplified model of four-area power system


Fig. 17 (b) Block diagram of four-area power system
The transfer function of area 1 of four area power with respect to disturbance input is obtained as

2.55s10  59.62 s 9  467.42 s8  1777.78s 7  3703.71s 6  4372.25s 5


2880.73s 4  1010.67 s 3  169.93s 2  11.212 s  0.1844
G4 ( s )  11 .
s  23.427 s10  184.495s 9  706.215s8  1487.132 s 7  1818.582 s 6
1215.317 s 5  452.731s 4  86.441s 3  7.727 s 2  0.2921s  0.00363
From Section 3.1, the predictive model is obtained as

N (s) 2.55s 2  39.099 s  1.2644


P4 m ( s )   3 .
D( s ) s  2.701s 2  1.258s  0.02489

The proposed controllers for LFC of four-area power system are obtained as
( s 3  2.701s 2  1.258s  0.02489)(0.0002 s 3  0.0096 s 2  0.16 s  1)
QD ( s )  and
(2.55s 2  39.099 s  1.2644)(0.04 s  1) 4

0.0002 s 6  0.01014 s 5  0.1862 s 4  1.444 s 3  2.903s 2  1.262 s  0.02489


GC ( s )  .
0.000006528s 6  0.0002427 s 5  0.002189 s 4  7.81e 5 s 3  2.193e 18 s 2

The proposed controller is implemented in each area of the four-area power system and 1% ELD

is applied in area 1 at t  1 sec. The simulation results are shown in Figs. 18(a)-18(d) for

frequency deviation and Figs. 18(e)-18(h) for tie-line power. It is observed from the figures that

the proposed scheme provides significantly better results than the conventional PID controller (

GC ( s )  9.9  15.9 / s  3.0 s. ) which is tuned by using Genetic algorithm [45].

(a) (b)

(c) (d)
(e) (f)

(g) (h)

Fig. 18 Responses of FD and TLP of four-area multi-unit multi-sources reheated HTPS with RPGs (a) Area-1 (b)
Area-2 (c) Area-3 (d) Area-4, and (e)- (h) Tie-line power for four area.

5. Conclusion

This article presents a novel dual loop-internal model control approach for LFC of single and

multi-area power systems. The proposed scheme utilizes a predictive model which is obtained by

model reduction and conventional IMC scheme. In order to establish the superiority of the

proposed scheme, the results are compared with existing techniques. Further, the robustness of

the proposed scheme is verified by inserting 50% parametric uncertainty and applying different

combinations of external load disturbances. The simulation results of a three-area reheated


hydro-thermal power system model with wind and solar energy penetrations and a four area

reheated hydro-thermal power system with renewable power sources such as wind power

generator, solar power generator, fuel cell and aqua-electrolyzer are also shown. It is observed

from the results that the proposed scheme gives a better disturbance rejection and fewer

oscillations. The proposed scheme can also be utilized for process control and biomedical

engineering applications.

References

[1] Kundur P. Power System Stability and Control. New York NY USA: Tata McGraw-Hill, 2012.
[2] Tan W. Tuning of PID load frequency controller for power systems. Energy Convers Manag
2009;50(6):1465–1472.
[3] Tan W. Unified tuning of PID load frequency controller for power systems via IMC. IEEE Tran Pow Sys
2010;25(1):341–350.
[4] Tan W. Load frequency control: problems and solutions. In: Proceedings of the 30th Chinese control
conference Yantai China 2011;6281–6286.
[5] Saxena S, Hote YV. Load frequency control in power systems via internal model control scheme and model-
order reduction. IEEE Tran Pow Sys 2013;28(3):2749–2757.
[6] Saxena S, Hote YV. Advances in Internal Model Control Technique: A Review and Future Prospects. IETE
Tech Review 2012;29(6):461–472.
[7] Saxena S, Hote YV. Decentralized PID load frequency control for perturbed multi area power systems. Electr
Power Energy Syst 2016;81:405–415.
[8] Liu T, Gao F. New insight into internal model control filter design for load disturbance rejection. IET Control
Theory Appl 2010;4(3):448–460.
[9] Xiao L, Gang C, Wang YG. IMC-PID Controller Design for Power Control Loop Based on Closed-loop
Identification in the Frequency Domain. In: Proceedings of the 14th IFAC Symposium on Large Scale
Complex Systems: Theory and Applications Riverside California USA IFAC 2016;49(4):79–84.
[10] Ghousiya BK, Seshagiri RA, Radhakrishnan TK. Enhanced IMC based PID controller design for non-
minimum phase (NMP) integrating processes with time delays. ISA Tran 2017;68:223–234.
[11] Sonker B, Kumar D, Samuel P. A modified two-degree of freedom-internal model control configuration for
load frequency control of a single area power system. Turk J Elec Eng Comp Sci 2017;25:4624 – 4635.
[12] Singh J, Chattterjee K, Vishwakarma CB. Two degree of freedom internal model control-PID design for LFC
of power systems via logarithmic approximations. ISA Tran 2018;72:185–196.
[13] Anwar N, Pan S. A new PID load frequency controller design method in frequency domain through direct
synthesis approach. Electr Power Energy Syst 2015;67:560–569.
[14] Shabani H, Vahidi B, Ebrahimpour M. A robust PID controller based on imperialist competitive algorithm for
load-frequency control of power systems. ISA Tran 2013;52:88–95.
[15] Padhan DG, Majhi S. A new control scheme for PID load frequency controller of single-area and multi-area
power systems. ISA Tran 2013;52:242–251.
[16] Sathya MR, Ansari MMT. Load frequency control using BAT inspired algorithm based dual mode gain
scheduled of PI controllers for interconnected power system. Electr Power Energy Syst 2015;64:365–374.
[17] Dash P, Saikia LC, Sinha N. Automatic generation control of multi area thermal system using Bat algorithm
optimized PD-PID cascade controller. Electr Power Energy Syst 2015;68:364–372.
[18] Rahman A, Saikia LC, Sinha N. Load frequency control of thermal system under deregulated environment
using biogeography based optimized three degree of freedom integral derivative controller. IET Gener
Transm Distrib 2015;9(15):2284–2293.
[19] Rahman A, Saikia LC, Sinha N. Automatic generation control of an unequal four area thermal system using
biogeography based optimized 3DOF-PID controller. IET Gener Transm Distrib 2016;10(16):1–31.
[20] Raju M, Saikia LC, Sinha N. Automatic generation control of a multi-area system using ant Lion optimizer
algorithm based PID plus second order derivative controller. Electr Power Energy Syst 2016;80:52–63.
[21] Sahu RK, Panda S, Rout UK, Sahoo DK. Teaching learning based optimization algorithm for automatic
generation control of power system using 2-DOF PID controller. Electr Power Energy Syst 2016;77:287–301.
[22] Sahu RK, Panda S, Biswal A, Sekhar GTC. Design and analysis of tilt integral derivative controller with filter
for load frequency control of multi-area interconnected power systems. ISA Tran 2016;16:251–264.
[23] Sondhi S, Hote YV. Fractional order PID controller for load frequency control. Energy Convers Manag 2014;
85:343–353.
[24] Zamani A, Barakati SM, Darmian SY. Design of a fractional order PID controller using GBMO algorithm for
load–frequency control with governor saturation consideration. ISA Tran 2016;64:56–66.
[25] Zheng Y, Zhou J, Xu Y, Zhang Y, Qian Z. A distributed model predictive control based load frequency
control scheme for multi-area interconnected power system using discrete-time Laguerre functions. ISA Tran
2017; 68:127–140.
[26] Mi Y, Fu Y, Li D, Wang C, Loh PC, Wang P. The sliding mode load frequency control for hybrid power
system based on disturbance observer. Int J of Electrical Power & Energy Systems 2016; 74: 446-452.
[27] Cui Y Xu L, Fei M, Shen Y. Observer based robust integral sliding mode load frequency control for wind
power systems. Control Engineering Practice 2017; 65: 1-10.
[28] Prakash S, Sinha SK. Load frequency control of multi-area power systems using Neuro-fuzzy hybrid
intelligent controllers. IETE J of Research 2015; 61: 526-532.
[29] Sivalingam R, Chinnamuthu S, Dash SS. A modified whale optimization algorithm-based adaptive fuzzy
logic PID controller for load frequency control of autonomous power generation systems. Automatika
2017;58:410-421.
[30] Arya Y, Kumar N. Design and analysis of BFOA-optimized fuzzy PI/PID controller for AGC of multi-area
traditional/restructured electrical power systems. Soft Computing 2017; 21: 6435–6452.
[31] Arya Y. Automatic generation control of two-area electrical power systems via optimal fuzzy classical
controller. Journal of the Franklin Inst 2018;355:2662–2688.
[32] Prakash S, Sinha SK. ALFC of hybrid multi-generation power system using UC and TCPS by ANFIS control
technique. Int J of Electro 2018:1-38.
[33] Prakash S, Sinha SK. Automatic load frequency control of six areas’ hybrid multi-generation power systems
using Neuro-fuzzy intelligent controller. IETE J of Research 2018;64: 471-481.
[34] Sonker B, Kumar D, Samuel P. Design of two degree of freedom-internal model control configuration for
load frequency control using model approximation. Int J Mod Sim 2018;1–11.
[35] Singh J, Vishwakarma CB, Chattterjee K. Biased reduction method by combining improved modified pole
clustering and improved Padé approximations. Appl Math Model 2016;40:1418–1426.
[36] Kishor N and Singh S. Simulated response of NN based identification and predictive control of hydro plant.
Expert Systems with Applications. 2007; 32: 233–244.
[37] Smith J, Fatehi F, Woods C. Transfer function identification in power system applications. IEEE Trans on
Power Systems. 1993; 8(3): 1282–1290.
[38] Li C, Zhou J, Li Q, An X, Xiang X. A new T–S fuzzy modelling approach to identify a boiler–turbine system.
Expert Systems with Applications. 2010; 37(3): 2214–2221.
[39] Liu Y, Wu QH, Zhou XX. Co-ordinated multi loop switching control of DFIG for resilience enhancement of
wind power penetrated power systems. IEEE Trans Sustain Energy 2016;7:1089–1099.
[40] Magdy G, Mohamed EA, Shabib G, et al. SMES based a new PID controller for frequency stability of a real
hybrid power system considering high wind power penetration. IET Renew Power Gener 2018;12:1304–1313.
[41] Dulal CD, Roy AK., Sinha N. GA based frequency controller for solar thermal–diesel–wind hybrid energy
generation/energy storage system. Electrical power and energy systems 2012;43:262–279.
[42] Shankar G, Mukherjee V. Load frequency control of an autonomous hybrid power system by quasi-
oppositional harmony search algorithm. Electrical power and energy systems 2016;78:715–734.
[43] Lee D, Wang L. Small-Signal Stability Analysis of an Autonomous Hybrid Renewable Energy Power
Generation/Energy Storage System Part I: Time-Domain Simulations. IEEE Transactions on Energy
Conversion 2008; 23(1): 311-320.
[44] Sanki P, Babu M. New approach in two-area interconnected AGC including various renewable energy sources
using PSO. Turk J Elec Eng & Comp Sci 2018; 26:1491-1504.
[45] Goldberg DE. Genetic algorithms. Pearson Education India, 2006.
Highlights:

1. A dual loop-internal model control scheme for load frequency control is proposed.

2. Single-area and multi-area power systems are considered for validation of scheme.

3. Model reduction is utilized to obtain the predictive model of proposed scheme.

4. Results with parametric uncertainties and external load disturbances are also presented.
Dual loop IMC structure for load frequency control issue of multi-
area multi-sources power systems
Bheem Sonker, Deepak Kumar, and Paulson Samuel
Department of Electrical Engineering, MNNIT Allahabad, Prayagraj, UP, India
*Correspondence: deepak_kumar@mnnit.ac.in

Abstract: In this paper, a dual loop-internal model control (DL-IMC) scheme for load frequency

control (LFC) issue of multi-area power systems is presented. The proposed scheme contains two

control loops for disturbance rejection and minimization of oscillations. The inner loop of the

proposed configuration contains an IMC controller and the outer loop controller is derived by

using a predictive model based on model reduction. In order to validate the proposed approach,

single and two area reheated thermal power system (TPS), and hydrothermal power systems

(HTPS) are considered. The robustness of the proposed scheme is shown by introducing 50%

perturbations in the system parameters for different values of external load disturbances. The

proposed method is extended for a three-area reheated HTPS with wind and solar power

penetrations along with random wind speed and solar irradiance. In addition to this, the

simulation results of a four-area reheated HTPS with renewable power sources such as wind

power generator, solar power generator, fuel cell and aqua-electrolyzer to show the effectiveness

of the proposed approach.

Keywords: internal model control; load frequency control; model reduction; multi-area power

systems.

Appendix A. Nomenclature

f , f1 , f 2 , Incremental frequency deviations TH 1 , TH 2 , Hydro turbine time


in Hz TH 3 , TH 4 ,
constants (s)
f3 , f 4
Pd , Pd 1, Pd 2 , External load disturbance in p.u. u1 ( s ), u2 ( s ) Control outputs of
MW controllers
Pd 3, Pd 4 ,
K p , K p1 , K p 2 , Electric system gains G p , G p1 , G p 2 , Transfer function of load
Hz/p.u.MW Gp3 , Gp 4 and machines
K p3 , K p 4
Tp , Tp 1 , Tp 2 Electric system time constants (s) Gg , Gg 1 , Gg 2 , Transfer function of steam
Gg 3 , Gg 4 governor
Tp 3 , Tp 4
Tt , Tt 1 , Tt 2 , Turbine time constants (s) Gr , Gr1 , Gr 2 , Transfer function of steam
Tt 3 , Tt 4 Gr 3 , Gr 4 reservoir

Tg , Tg 1 , Tg 2 , Governor time constants (s) Gt , Gt1 , Gt 2 , Transfer function of steam


Gt 3 , Gt 4 turbine
Tg 3 , Tg 4 ,
R, R1 , R2 , Speed regulations due to GHg1 , GHg 2 , Transfer function of hydro
R3, R4 , R5, governor action in Hz/p.u. MW GHg 3 , GHg 4 governor

R6 , R7, R8 ,
Tr , Tr1 , Tr 2 , Reheated time constants (s) GHR1 , GHR 2 , Transfer function of hydro
Tr 3 , Tr 4 GHR 3 , GHR 4 , reservoir

K r , K r1 , K r 2 , High pressure turbine power GHw1 , GHw 2 , Transfer function of hydro


Kr3 , Kr 4 fraction GHw3 , GHw 4 turbine

Tw1 , Tw 2 , Water time constants (s) PTie Tie-line power error


Tw3 , Tw 4
THg 1 , THg 2 , Hydro governor time constants (s) T12 , T23 , Synchronizing coefficients
THg 3 , THg 4 T34 , T41
THR1 , THR 2 , Hydro reservoir time constants (s) Pr1 , Pr 2 , Power system rated area
THR 3 , THR 4 Pr 3 , Pr 4 capacity (MW)
K wpg1 , K wpg 2 , Wind power generator (WPG) K spg1 , K spg 2 , Solar power generator
gains (SPG) gains
K wpg 3 , K wpg 4 K spg 3 , K spg 4
K fc1 , K fc 2 , Fuel cell (FC) gains K ae1 , K ae 2 , Aqua- electrolyzer (AE)
K ae3 , K ae 4 gains
K fc 3 , K fc 4
Twpg1 , Twpg 2 , WPG time constants (s) Tspg1 , Tspg 2 , SPG time constants (s)
Twpg 3 , Twpg 4 Tspg 3 , Tspg 4
T fc1 , T fc 2 , FC time constants (s) Tae1 , Tae 2 , AE time constants (s)
T fc 3 , T fc 4 Tae 3 , Tae 4

1. Introduction
During operation of a large-scale power system, the fluctuations caused by load demand

may affect the system frequency and tie-line power (TLP) interchange considerably. Therefore, a

load frequency controller is required to maintain the frequency and TLP exchange at the

scheduled values. In general, a load frequency controller maintains the performance of a power

system during external load disturbance (ELD) and parametric uncertainties [1]. Since the last

few decades, several approaches using different control strategies have been proposed for LFC of

single area and multi-area power systems (MAPS) during normal operating conditions and load

perturbations. The IMC is one of the popular approaches for LFC of single area and MAPS.

Using two degrees of freedom-internal model control (TDF-IMC) scheme, Tan [2-3] designed a

proportional-integral-derivative (PID) controller for LFC. Further, PID controller based

problems and solutions for LFC are presented in [4]. Later, Saxena and Hote [5] presented a

TDF-IMC based approach for LFC of single area non-reheated TPS. Various structures of TDF-

IMC for eliminating load disturbance and its applications can be referred from [6]. A PID

controller is designed for LFC of MAPS in [7] and a modified IMC is presented in [8] for a

better performance of the closed-loop system. Xiao et al. [9] presented an IMC based PID

controller for power control in the frequency domain. Another IMC based PID controller is

introduced by Ghousiya et al. [10] using an optimal H2 minimization. A modified TDF-IMC

scheme for LFC of a non-reheated TPS using model reduction (MR) is presented in [11].

Recently, Singh et al. [12] presented a TDF-IMC based PID controller for LFC of the single area

and multi-area reheated HTPS via logarithmic approximations. The PID controllers for LFC

have also been designed by direct synthesis approach using frequency response matching,

imperialist competitive algorithm and Laurent series expansion in [13], [14] and [15]. In addition

to this, Sathya et al. [16] designed PI controllers with dual mode gain scheduling and Dash et al.
[17] presented PD-PID controller using BAT inspired algorithm for an interconnected power

system. Rahman et al. [18] introduced an optimized 3DOF-PID controller using the

biogeography-optimized technique for LFC of a TPS under a deregulated environment. Later,

Rahman et al. [19] presented another approach for automatic generation control (AGC) of

unequal four-area TPS. In [20], a PID plus second order derivative controller for AGC of MAPS

is designed using the ant-lion optimizer algorithm. Subsequently, Sahu et al. [21-22] designed

TDF-PID controllers using optimization algorithm based on teaching-learning for AGC of a

power system and designed tilt integral derivative controller with a filter for LFC of MAPS.

Furthermore, LFC using fractional order-PID controllers are presented in [23], [24] and a

distributed model predictive control based LFC for MAPS using discrete-time Laguerre

functions is presented in [25]. The observer-based sliding mode control approaches for load

frequency control are presented in [26] and [27]. An artificial neural network and fuzzy-based

adaptive neuro-fuzzy inference system (ANFIS) approach for AGC of three area HTPS is

presented in [28]. Some other fuzzy based approaches for load frequency control are presented

using a modified whale optimization in [29], bacterial foraging optimization in [30], optimal

controller in [31], ultra-capacitor and thyristor control phase shifter in [32] and a neuro-fuzzy

intelligent controller in [33].

It is well known that a multi-area multi-source power system results into a higher order

mathematical model which poses several issues in designing of a suitable load frequency

controller. Such controllers give a sluggish response and require more computational effort.

Therefore, it is desired to use a reduced order mathematical model to design controller such that

improved performance may be obtained. Although several controllers using IMC have been

presented for LFC problem, there is a large scope for improving the performance with
disturbance rejection and load changes issue. Therefore, this article presents the DL-IMC scheme

for LFC of MAPS as an extension of recent work [34] by employing a MR method [35]. The

contributions of the paper are as follows:

1) A reduced order model of the original power system model is obtained which is used as a
predictive model in the proposed approach.
2) A controller is developed by incorporating pole clustering and improved Padѐ

approximation [35] to the proposed scheme for achieving less computational effort.

3) The robustness of the proposed approach is verified by introducing 50% variations in

the parameters of the power system model.

4) Performance indices such as integral square error (ISE), integral absolute error (IAE) and

integral time absolute error (ITAE) are computed with original and perturbed system

parameters for a comparative analysis of the proposed approach.

To show the effectiveness of the proposed method, it is compared with existing methods [3,

13, 15, 23, 30] for single area reheated TPS and with [12, 30] for single area and two area

HTPS. It is observed that the proposed scheme minimizes the frequency deviations and TLP

exchange better than the existing techniques. Further, the proposed scheme also maintains the

robustness under random load variations which is shown by introducing different values of

ELD as well as parametric uncertainties. The article is arranged as follows: a brief

description of single area reheated TPS, and single and two area reheated HTPS are included

in Section 2 whereas the proposed scheme is explained in Section 3. The simulation results of

the proposed scheme with nominal and random load changes are discussed in Section 4.

Finally, the conclusion is presented in Section 5.

2. Power system models under consideration


In this section, single area and multi-area multi-source power system models have been

considered for validation of the proposed approach. A single area TPS [1, 3, 13, 15, 23] with a

reheated turbine is considered as a first system model as shown in Fig. 1. The mathematical

model of a single area reheated TPS consists of linearized models of the steam governor, steam

reservoir, reheated turbine, and load and machines. A reheated turbine provides higher thermal

efficiency and the transfer function of a reheated turbine involves an additional time constant Tr1

along with the turbine time constant. The second considered model is a two area TPS [30] which

has reheated turbine as shown in Fig. 2. The third model is a single area HTPS [12] which has

multi-source generation with reheated and hydro turbines as shown in Fig. 3. The mathematical

model of a single area reheated TPS consists of linearized models of the steam governor, steam

reservoir, reheated turbine, hydro governor, hydro reservoir, hydro turbine along with load and

machines. The fourth model is a two area multi-unit multi-source HTPS [12] with reheated and

hydro turbines in each area of the model as shown in Fig. 4.

Pd
_
u+ 1 (1  sK r1Tr1 ) 1 +
Kp f
_ (1  sTg ) (1  sTr1 ) (1  sTt ) (1  sTp )
Gg Gr Gt Gp
1
R

Fig. 1. Single area reheated TPS


1
B1 R1

Pd 1
+ ACE1 _ _
+ 1 (1  sK r1Tr1 ) 1 + K p1 f1
Controller
+ (1  sTg1 ) (1  sTr1 ) (1  sTt1 ) _ (1  sT p1 )
Gg1 Gr1 Gt1 G p1

+
T12
s _

+ ACE 2 _
+ 1 (1  sK r 2Tr 2 ) 1 + K p2 f 2
Controlller
_ _ (1  sTg 2 ) (1  sTr 2 ) (1  sTt 2 ) _ (1  sT p 2 )
Gr 2 Gg 2 Gt 2 Gp2

1 Pd 2
B2
R2

Fig. 2 Two area reheated TPS

1 (1  sK r1Tr1 ) 1 Pd 1
(1  sTg 1 ) (1  sTr1 ) (1  sTt1 ) _
u1 + Gg 1 Gr1 Gt1 + K p1 f1
_ + (1  sT p1 )
+
1 (1  sTHR1 ) (1  sTw1 )
G p1
_ (1  sTHg 1 ) (1  sTH 1 ) (1  0.5sTw1 )
1 GHg 1 GHR1 GHw1
R1 1
R2

Fig. 3. Single area reheated HTPS


1
R1 1
B1
R2
1 (1  sK r1Tr1 ) 1 Pd 1
+ ACE1 _ (1  sTg 1 ) (1  sTr1 ) (1  sTt1 ) _
Controller
+ Gg 1 Gr1 Gt1 + K p1 f1
+ _ + _ (1  sT p1 )
+
1 (1  sTHR1 ) (1  sTw1 )
G p1
(1  sTHg 1 ) (1  sTH 1 ) (1  0.5sTw1 )
GHg 1 GHR1 GHw1
+
T12
s _
_
_

1 (1  sK r 2Tr 2 ) 1
+ ACE 2 (1  sTg 2 ) (1  sTr 2 ) (1  sTt 2 ) _
+ Gg 2 Gr 2 Gt 2 + K p2 f 2
Controlller
_ _ + _ (1  sT p 2 )
+
1 (1  sTHR 2 ) (1  sTw 2 )
Gp2
_ (1  sTHg 2 ) (1  sTH 2 ) (1  0.5sTw 2 )
1 GHg 2 GHR 2 GHw 2 Pd 2
B2 1
R3
R4

Fig. 4. Two area reheated HTPS

The transfer functions (TFs) of each block (refer Appendix A for the values of parameters) are

shown in the figures. The overall TF of reheated TPS G ( s ) (refer Fig. 1) with respect to load

disturbance as reference input is given by

f Gp  s
G  s   (1)
Pd 1  G s G s G ( s )G s 1
p  g   r t 
R

It is clear that the LFC is a load disturbance rejection problem which uses the feedback law

u( s )  Gc ( s )f ( s ) to stabilize the power plant G ( s ) under load disturbance Pd ( s ) and

reduces the effect of Pd ( s).

The TF of area 1 for the two area reheated TPS (refer Fig. 2), is obtained with respect to

external load disturbance as reference input which is given by


f1 G p1 ( s ) B1
G2 rtps ( s )   (2)
Pd 1 1  G ( s )G ( s )G ( s )G ( s ) 1
p1 g1 r1 t1
R1

The TF of single area reheated HTPS (refer Fig. 3) is derived with respect to external load

disturbance as reference input which is as below

f1 G p1 ( s )
G1 ( s )   . (3)
Pd 1 1  G p1 ( s )((Gg1 ( s )Gr1 ( s )Gt1 ( s ) / R1 )  (GHg1 ( s )GHR1 ( s )GHw1 ( s ) / R2 ))

For the two-area case, the area control errors and TLP expressions are written as

e1 (t )  ACE1  B1f1  PTie ,


e2 (t )  ACE2  B2 f 2  PTie , (4)
T12
PTie  (f1  f 2 ).
s
where B1 and B2 are the frequency bias constants (p.u.MW/Hz), ACE1 and ACE2 are the area

control errors of area 1 and area 2 respectively. The area controls errors are used to regulate the

output power of the generator as per the ELD. From Eq. (3), it is clear that the LFC is the main

issue against RLD and variation in ELD. The feedback control law for area 1 and 2 are obtained

as follows

u1 ( s )  GC1 ( s ) ACE1 ,
(5)
u2 ( s )  GC 2 ( s ) ACE2 ,

where GC1 ( s ) and GC 2 ( s ) are the feedback controllers. In two area case, a decentralized

controller can be obtained by assuming that there is no TLP exchange i.e. PTie  0. Based on

the above equations, the TF of area 1 for the two area reheated HTPS (refer Fig. 4), is obtained

with respect to external load disturbance as reference input as

f1 G p1 ( s ) B1
G2 ( s )   , (6)
Pd 1 1  G p1 ( s )((Gg1 ( s )Gr1 ( s )Gt1 ( s ) / R1 )  (GHg1 ( s )GHR1 ( s )GHw1 ( s ) / R2 ))

Thus, the controllers for LFC of each area can be obtained similarly.
3. Proposed scheme

The proposed DL-IMC scheme is designed to reduce the effects of ELD and parametric

uncertainties during operation of a MAPS. The control block diagram of the proposed scheme is

shown in Fig. 5(a) which consists of a plant model G ( s ), a disturbance-rejection controller

(DRC) QD ( s ) in the inner loop (IL), the PM Pm ( s ) in series with QD ( s ) in the forward path, and

an outer loop controller (OLC) GC ( s ) in negative feedback path. In order to simplify the control

block diagram, the summing junction before QD ( s ) can be separated in two parts. The first part

gives a cascade connection of Pm ( s ) and QD ( s ) resulting into the forward path controller (FPC)

K m ( s )  Pm ( s )QD ( s ) as shown in Fig 5(b) that is a simplified version of Fig. 5(a). The second

part consists of unity gain in forward path and QD ( s ) in feedback path which gives

K d ( s )  1 / (1  QD ( s )) as the series controller (SC). The OLC is incorporated in the system for

removing ELD which causes the system output to become uncontrolled. In order to stabilize the

uncontrolled response of the system caused by OLC, an FPC is incorporated in the proposed

approach. However, the FPC produces oscillations in the system, which is minimized by the

addition of an IL controller that also helps in fast removal of ELDs. Further, a suitable model

reduction method is necessary to obtain the predictive model Pm (s ) which is used to design

QD (s ) and GC ( s ). The tuning parameter  d used for design of QD (s ) must be chosen properly.

As the design of OLC is based on the TDF-IMC scheme, it gives faster rejection of external load

disturbances and the inner loop controller reduces the overall gain of the closed loop system.
Pd (s )
_ Pd (s )
R (s ) + + C (s ) _
G(s )
_ _ R (s ) + + + C (s )
G(s ) K d (s )
_ _
Pm (s ) QD (s )
K m (s )
+ +

GC (s ) GC (s )

Fig. 5. a) Proposed DL-IMC scheme b) Simplified version of proposed scheme

The PM and conventional IMC configurations are used for developing the DL-IMC

scheme and the PM is derived by using an MR technique [35] which is discussed as below:

3.1 Derivation of the PM

A generalized k th order TF G ( s ) for single area reheated TPS, single area HTPS and each

of two areas HTPS is considered as

g 0  g1s  .....  s k 1 g k 1
G (s)  . (7)
h0  h1s  h2 s 2  ....  s k hk

The PM Pm ( s )  N ( s ) D( s ), of the proposed scheme is derived from the system model G ( s ) by

using pole clustering and improved Padѐ approximation [35] for the denominator and numerator

of PM respectively. The steps to obtain D( s ) are as follows:

1. Sort the real poles i.e. p1  p2  p3 ......  pv and complex poles ( 1  j1 ,  2  j 2 …..

 v 1  j v 1 ) i.e. 1   2  .......   v 1 and 1   2  .......   v 1 of G ( s ).

2. Set   1;

3. Compute the pole cluster centers (PLC) of real poles and complex poles as

1 1 1
 v  1    v 1  1    v 1  1  
 j       v  , Aej       (v  1)  and Bej       (v  1)  . (8)
 i 0  pi    i 0  i    i 0   i  
4. Set     1;

5. Generate another PLC as

1 1 1
  1 1     1 1     1 1  
       2  , Ae      2 and Be    *   2  . (9)
 p1  1  
   1 L 1  
   1 L 1  
 

6. If   v  1 for real pole and   v for complex poles, then go to step-7. Otherwise, go to

step 4.

7. Finally, the cluster centers are obtained as pe1  v 1 for real pole, e*1  Aev  jBev and

e1  Aev  jBev for complex poles. Therefore, the denominator of PM is obtained as

D ( s )  ( s  pe1 )( s  e*1 )( s  e1 )  x0  x1s  x2 s 2  x3s 3 (10)

Further, N ( s ) is determined by matching time moments and Markov parameters of G ( s ) and

PM is obtained by expanding G ( s ) as below:


G ( s )   M i s  i 1 (about s  )
i 0

(11)
   Ti s i
(about s  0),
i 0

where Ti and M i are the i th Time moment and Markov parameters of G ( s ) respectively. If

N ( s )  n0  n1s  n2 s 2 , then

g 0  h0T0 ,
g1  h0T1  h1T0 ,
n0  x0T0 , (12)
n1  x0T1  x1T0 ,
n2  x3 M 0 .

Therefore, the PM is obtained as

N ( s) n0  n1s  n2 s 2
Pm ( s )   . (13)
D ( s ) x0  x1s  x2 s 2  x3s 3
3.2 Design of controllers

The controllers of the proposed DL-IMC scheme are designed by using the conventional

IMC configuration [3, 5] which is shown as below:

1. Factorize Pm ( s ) as

Pm ( s )  Pm  ( s ) Pm  ( s ), (14)

where Pm  ( s ) and Pm  ( s ) are the minimum-phase and non-minimum-phase parts.

2. Design the IMC controller QD ( s ) as

QD ( s )  Pm1 ( s )Qd ( s ), (15)

1  1s   2 s 2   3s 3
where Qd ( s )  is the third order low pass filter,  d is the tuning
( d s  1) n 1

parameter and n is the total number of poles Pm ( s ) that are p1 , p2 ,  , pn . The values of

1,  2, and  3 are obtained by

1  1s   2 s 2   3s 3
(1  ) |s  p1, p2, .... pn  0. (16)
( d s  1) n 1

3. The design of OLC GC ( s ) is based on the complementary sensitivity transfer function

T ( s ) of the IMC feedback configuration [8] which is given by

T ( s )  Pm ( s )QD ( s )
 Pm ( s ) Pm1 ( s )Qd ( s )
Pm  ( s )(  3s 3   2 s 2  1s  1) (17)
 .
( d s  1) n 1
P ( s )GC ( s )
As T ( s )  m , we get GC ( s ) from (15) as
1  Pm ( s )GC ( s )

Pm1 ( s )(  3s 3   2 s 2  1s  1)
GC ( s )  . (18)
{( d s  1) n 1  Pm  ( s )(  3s 3   2 s 2  1s  1)}
4. Simulation Results

The performance and robustness of the proposed method for LFC of single area reheated

TPS, single and two areas reheated HTPS are discussed in this section by varying ELD.

4.1. Single-area reheated TPS

A single-area reheated TPS [3] is considered as shown in Fig. 1 with the following system

parameters:

K p  120, Tp  20, Tt  0.3, Tg  0.08, R  2.4, Tr  4.2, and K r  0.35.

After substituting the values in Eq. (1), the TF of system G ( s ) is obtained as

6s 3  96.42 s 2  272.62 s  59.53


G( s)  (19)
s 4  16.12 s 3  46.42 s 2  48.65s  25.3

The poles of system are 12.79, 2, 0.665  0.7395i. As discussed in Section 3.2, the controllers

QD ( s ) and GC1 ( s ) are designed by using minimum phase Pm  ( s ) of the PM Pm ( s ) where the

Pm ( s ) is obtained by using Section 3.1 as

N ( s) 6 s 2  23.147 s  5.21
Pm ( s )   3 (20)
D ( s ) s  3.566 s 2  3.963s  2.211

The step responses of G ( s ) and Pm ( s ) PM are compared as shown in Fig. 6. It is clear from

the figure that the response of predictive model completely follows the path of G ( s ) and it also

achieves approximately zero steady-state error. Therefore, this predictive model can be used for

designing the controllers.

Remark 1: The model reduction method must be chosen such that the resulting reduced model

matches the steady state response of the original model. Further, the reduced model must be

stable.
Step Response
6

Amplitude
3

System model
1
Predictive model
0
0 2 4 6 8 10 12
Time (seconds)

Fig. 6. Comparison of step responses between system model and predictive model

After substituting, the values  d  0.05 and n  3 in Eq. (15) and Eq. (18) respectively, the

DRC QD ( s ) is obtained as

( s 3  3.566 s 2  3.963s  2.211)(0.0005s 3  0.015s 2  0.2 s  1)


QD ( s )  , (21)
(6 s 2  23.147 s  5.21)(0.05s  1) 4

where  3  0.0005,  2  0.015 and 1  0.2 .

Finally, OLC GC ( s ) is obtained as

0.0005s 6  0.01678s 5  0.2555s 4  1.774 s 3  4.392 s 2  4.405s  2.211


GC ( s )  . (22)
0.000037 s 6  0.0001447 s 5  0.00003256s 4  8.087e  7 s 3  1.808e 17 s 2

An ELD Pd  0.01 is applied at t  2 sec. to the system for determining performance of

the proposed method. Further, the robustness of the proposed method is analysed by inserting

50% variations in the parameters of system. Hence, the parameters of single area reheated TPS

are expressed as

K p  [60,180], Tp  [10,30], Tt  [0.15, 0.45], Tg  [0.04, 0.12], Tr  [2.1, 6.3],


K r  [0.175, 0.525] and R  [1.2,3.6];

The responses of the proposed scheme are shown in Figs. 7(a)-7(c) for nominal value

(NV), lower bound (LB) and upper bound (UB) of the system parameters. The proposed method
is compared with different controllers such as fractional order PID [23], IMC-PID [3], PID [13,

15]. It is clear from the figures that the proposed method removes ELDs with fewer oscillations

and settles much quicker than the existing techniques [3, 13, 15, 23]. The performance of the

proposed scheme and existing techniques are also compared through the error indices for

nominal and perturbed parameters. The expressions for the error indices are given as follows:

(a) ISE   e(t ) dt.


2

0

(b) IAE   e(t ) dt.


0

(c) ITAE   t e(t ) dt.


0

It is observed from Tables 1, 2 and 3 that the performance indices of the proposed scheme are

significantly lower than the existing techniques [3, 13, 15, 23]. Therefore, the proposed DL-IMC

scheme effectively handles the frequency deviations and rejects the disturbances very well

during power system operation.

Remark 2: The tuning parameter  d required in the design of DRC needs to be chosen by taking

a suitable compromise between closed-loop response and robustness. A large value of tuning

parameter results into a sluggish closed-loop response and appropriate robustness towards the

variation in the system parameters. On the other hand, a smaller value generates a comparatively

faster closed-loop response and inappropriate robustness.

Remark 3: A suitable optimization algorithm can also be used to obtain the tuning parameters

used in the inner and outer loop controllers to improve the performance of the proposed

approach. However, it is avoided in the current work to simplify the proposed method.
Remark 4: It is to be noted that the proposed approach can also be used for other stable
processes and it works well. In the case of unstable processes, an additional compensator may be
required.

(a) (b)

(c)
Fig. 7. Frequency deviations (FD) of the proposed scheme for single area reheated TPS of 1% load variation with (a)

NV, (b) LB and (c) UB.

Table 1
Single area reheated TPS for nominal value

Error indices
Methods 1% Load variation
ISE IAE ITAE
Proposed method 1.436×10-8 5.962×10-5 4.678×10-4
Tan’s method [3] 6.152×10-5 1.118×10-2 7.419×10-2
Anwar & Pan’s [13] 6.938×10-6 3.992×10-3 2.672×10-2
Padhan & Majhi’s [15] 2.002×10-5 5.722×10-3 3.972×10-2
Sondhi & Hote’s [23] 1.594×10-5 3.937×10-3 2.883×10-2

Table 2
Single area reheated TPS for LB

Error indices
Methods 1% Load variation
ISE IAE ITAE
Proposed method 1.377×10-8 5.873×10-5 4.606×10-4
Tan’s method [3] 4.823×10-5 7.828×10-3 5.597×10-2
Anwar & Pan’s [13] 5.466×10-6 3.636×10-3 2.284×10-2
Padhan & Majhi’s [15] 1.595×10-5 4.959×10-3 3.334×10-2
Sondhi & Hote’s [23] 1.476×10-5 4.199×10-3 2.938×10-2

Table 3
Single area reheated TPS for UB

Error indices
Methods 1% Load variation
ISE IAE ITAE
Proposed method 1.449×10-8 6.005×10-5 4.71×10-4
Tan’s method [3] 5.746×10-5 1.05810-3 7.043×10-2
Anwar & Pan’s [13] 7.572×10-6 4.244×10-3 2.683×10-2
Padhan & Majhi’s [15] 2.356×10-5 6.231×10-3 4.2×10-2
Sondhi & Hote’s [23] 1.618×10-5 3.74×10-3 2.74×10-2

4.2. Two-area reheated TPS extension

The proposed method is implemented for two-area reheated TPS power system [30] as

shown in Fig. 2. The values of parameters of system are given as below:

K p1  K P 2  120, Tp1  Tp 2  20, Tt1  Tt 2  0.3, Tg1  Tg 2  0.08, Tr1  Tr 2  10, K r1 


K r 2  0.5, R1  R2  2.4, B1  B2  0.425, T12  T21  0.545, Pr1  Pr 2  2000.

After substituting the values in Eq. (2), we obtain the TF models of area 1 and area 2 of two

areas reheated TPS. The transfer function G2 rtps ( s ) of area 1 of two area reheated TPS with

respect to disturbance input is obtained as


2.55s 3  40.625s 2  110.3s  10.625
G2 rtps ( s )  (23)
s 4  15.94 s 3  40.042 s 2  58.42 s  10.625

and the predictive model is obtained by using Section 3.1 as

N (s) 2.55s 2  4.903s  0.4702


Pm 2 rtps ( s )   3 (24)
D( s ) s  3.023s 2  2.607 s  0.4702

The proposed controllers for LFC of two-area reheated TPS power system are obtained as

( s 3  3.023s 2  2.607 s  0.4702)(0.0002 s 3  0.0092 s 2  0.16 s  1)


QD ( s )  and
(2.55s 2  4.903s  0.4702)(0.04 s  1) 4

0.0002 s 6  0.009805s 5  0.1883s 4  1.508s 3  3.444 s 2  2.682 s  0.4702


GC ( s )  .
0.000006528s 6  0.0001554 s 5  0.001296 s 4  0.001988s 3  0.0001881s 2

A 1% ELD of full load is applied at t  2 sec. to area 1 of the system and the performance of

proposed method is analysed by introducing 50% parametric uncertainties as

K p1  K p 2  [60,180], Tp1  Tp 2  [10,30], Tt1  Tt 2  [0.15, 0.45], Tg1  Tg 2  [0.04, 0.12],


Tr1  Tr 2  [5,15], K r1  K r 2  [0.25, 0.75], and R1  R2  [1.2,3.6].

The simulation results are shown in Figs.8 (a)-8(c) for NV, LB and UB and it is observed from
the figures that the proposed scheme provides significantly better results than the recently
proposed fuzzy PI controller [30].

(a) (b)
(c)
Fig. 8. (a) FD of area 1, with 1% load variations in area 1for NV, LB and UB (b) FD of area 2, with 1% load
variations in area1 for NV, LB and UB, (c) TLP exchange with 1% load variations in area 1 for NV, LB and UB.

4.3. Single-area reheated HTPS

The values of parameters for single-area reheated HTPS (refer Fig. 3) are considered from

[12] as

K p1  120, Tp1  20, Tt1  0.3, Tg1  0.08, Tr1  10, K r1  0.5, Tw1  1, THg1  41.6, THR1  5,
TH 1  0.513, R1  R2  2.4.

After substituting the values in Eq. (3), the TF model G1 ( s ) obtained as

6s 6  119.4 s 5  663.3s 4  1438s 3  1145s 2  124.2 s  2.343


G1 ( s )  7 . (25)
s  19.96s 6  109.2 s 5  262.1s 4  352.6s 3  364.8s 2  66.83s  1.972

As discussed in Section 3.2, QD1 ( s ) and GC1 ( s ) are determined by using Pm1 ( s ) that is obtained

from Pm1 ( s ) . Hence, from the approach given in Section 3.1, the 3rd order PM Pm1 ( s ) is obtained

as

N ( s) 0.05116  2.3065s  6s 2
Pm1 ( s )   . (26)
D ( s ) 0.04306  1.118s  0.8763s 2  s 3
A comparison of step responses of single area reheated HTPS G1 ( s ), PM Pm1 ( s ) , and the

reduced models by Singh et al. [12] is shown in Fig. 9 and it is observed that the utilized MR

approach produces a satisfactory PM.

Step Response
6
G1(s): System model
5 Pm1(s): Predictive model
R1 [12]: Singh et al.
R2 [12]: Singh et al.
4
Amplitude

0
0 5 10 15 20
Time (seconds)
Fig. 9. Comparison of step responses between the original system and reduced models

After substituting, the values  d  0.1 and n  3 in Eq. (15) and Eq. (16) respectively, the

DRC QD1 ( s ) is obtained as

( s 3  0.8763s 2  1.118s  0.04306)(0.0039 s 3  0.0599 s 2  0.4 s  1)


QD 1 ( s )  , (27)
(6 s 2  2.3065s  0.05116)(0.1s  1) 4
where  3  0.0039,  2  0.0599 and 1  0.4 .

Finally, OLC GC 1 ( s ) is obtained as

0.0039 s 6  0.06332 s 5  0.4569 s 4  1.418s 3  1.326s 2  1.135s  0.04306


GC1 ( s )  . (28)
0.0006s 6  0.0008307 s 5  0.0008358s 4  0.0002358s 3  5.116e 06 s 2

A 1% load variation in full load i.e. 2000MW at t  2.0 sec. is applied to the system and

the robustness of the proposed method is analysed by inserting 50% parametric uncertainties as

K p  [60,180], Tp  [10,30], Tt1  [0.15, 0.45], Tg1  [0.04, 0.12], Tr1  [5,15],
K r1  [0.25, 0.75], Tw1  [0.5,1.5], THg1  [20.8, 62.4], THR1  [2.5, 7.5], TH 1  [0.2565, 0.7695]
and R1  [1.2,3.6].
The corresponding results are shown in Figs. 10(a)-10 (c) for NV, LB and UB. It is

observed from the results that ELD rejection is very fast and smoother as compared to Singh et

al.’s technique [12].

(a) (b)

(c)
Fig. 10. Frequency deviations (FD) of the proposed scheme for single area reheated HTPS with (a) NV (b) LB and

(c) UB.

The performance indices of the proposed method are expressed in Tables 4, and it is clear

that the proposed method gives significantly lower values than the existing techniques [12].

Therefore, the proposed DL-IMC rejects the disturbances very fast and maintains the robustness.

Table 4
Single area reheated HTPS
Error 1% Load variation
indices NV LB UP NV LB UP
Singh et al. IMC-PID [12] Proposed method
ISE 1.604×10-5 1.948×10-5 1.838×10-5 2.823×10-8 2.664×10-8 2.914×10-8
IAE 3.939×10 -3 5.145×10 -3 3.969×10-3 1.029×10-4 1.007×10-4 1.05×10-4
ITAE 2.0×10-2 3.564×10-2 2.879×10-2 5.886×10-4 7.759×10-4 8.103×10-4

4.4. Two-area reheated HTPS

The values of parameters for two area reheated HTPS (refer Fig. 4) are taken from [12] as

K p1  K P 2  120, Tp1  Tp 2  20, Tt1  Tt 2  0.3, Tg1  Tg 2  0.08, Tr1  Tr 2  10, K r1 


K r 2  0.5, Tw1  Tw 2  1, THg1  THg 2  41.6, THR1  THR 2  5, TH 1  TH 2  0.513, R1  R2
 R3  R4  2.4, B1  B2  0.425, T12  T21  0.545, Pr1  Pr 2  2000.

After substituting the values in Eq. (6), we obtain the TF models of area 1 and area 2 of two

areas reheated HTPS. The analysis of only one area is shown as we know that both TFs come out

to be equal for the same values of parameters. The TF model G2 ( s ) of area 1 is obtained as

2.55s 6  50.74 s 5  281.9 s 4  611.1s 3  486.6s 2  52.8s  0.99


G2 ( s )  . (29)
s 7  19.96s 6  109.2 s 5  262.1s 4  352.6s 3  364.8s 2  66.83s  1.972

From Section 3.2, the PM Pm 2 ( s ) is obtained as

N ( s) 0.02162  0.9816 s  2.55s 2


Pm 2 ( s )   . (30)
D ( s ) 0.04306  1.118s  0.8763s 2  s 3

Similar to the previous case, after substituting, the values  d  0.2 and n  3 in Eq. (15)

and Eq. (17) respectively, the DRC QD 2 ( s ) is obtained as

( s 3  0.8763s 2  1.118s  0.04306)(0.0306 s 3  0.2382 s 2  0.7999 s  1)


QD 2 ( s )  , (31)
(2.55s 2  0.9816 s  0.02162)(0.2 s  1) 4
where  3  0.0306,  2  0.2382 and 1  0.7999.

Finally, OLC GC 2 ( s ) for two area case is obtained as


0.0306s 6  0.265s 5  1.043s 4  1.969 s 3  1.781s 2  1.152 s  0.04306
GC 2 ( s )  . (32)
0.00408s 6  0.005141s 5  0.005999 s 4  0.002052 s 3  0.0001371s 2  2.162e 06 s

To check the robustness of the proposed scheme, the parametric uncertainties are considered for

area 1 and area 2 and the responses of the proposed scheme are obtained as follows:

Case 1. The ELD of 2% is applied at t  2 sec. to area 1 of the system for NV and the

simulation results are shown in Figs. 11(a)-11(c). It is observed from the figures that the

proposed scheme provides the better results than the methods used in [12].

(a) (b)

(c)

Fig. 11. (a) FD of area-1 with 2% in area 1 for NV, (b) FD of area-2 with 2% load variation in area 1 for NV, and (c)

TLP exchange with 2% load variation in area 1 for NV.


Case 2. A 2% ELD of full load is applied at t  2 sec. to area 1 of the system and the robustness

of the proposed scheme is checked by inserting 50% parametric uncertainties as

K p1  K p 2  [60,180], Tp1  Tp 2  [10,30], Tt1  Tt 2  [0.15, 0.45], Tg1  Tg 2  [0.04, 0.12],


Tr1  Tr 2  [5,15], K r1  K r 2  [0.25, 0.75], Tw1  Tw 2  [0.5,1.5],
THg1  THg 2  [20.8, 62.4], THR1  THR 2  [2.5, 7.5], TH 1  TH 2  [0.2565, 0.7695]
and R1  R2  R3  R4  [1.2,3.6].

It is observed from the Figs. 12(a)-12(c) for LB and Figs. 13(a)-13(c) for UB that the proposed

scheme gives better disturbance rejection than [12].

(a) (b)

(c)
Fig. 12. (a) FD of area 1, with 2% load variations in area 1for LB, (b) FD of area 2, with 2% load variations in area1

for LB, (c) TLP exchange with 2% load variations in area 1 for LB.
(a) (b)

(c)

Fig. 13. (a) FD of area 1, with 2% load variations in area 1for UB, (b) FD of area 2, with 2% load variations in area1

for UB, (c) TLP exchange with 2% load variations in area 1 for UB.

The performance of the proposed scheme and existing techniques are compared through

the error indices for nominal parameters and parametric uncertainties of two-area reheated

HTPS. It is observed from Tables 5, 6 and 7 that the performance indices of the proposed scheme

are significantly lower than the existing techniques [12]. Therefore, the proposed scheme gives

better disturbance rejection and robustness. The tie-line powers of two area reheated HTPS by

using proposed and existing methods [12] are shown in Table 8 and it is clear that the proposed

method reduces the tie-line power exchange as compared to [12].


Case 3. The load change of amplitude 0.15, -0.2, 0.2, and 0.1 are applied at t= 5, 15, 25, and 35

sec. to area 1 for analyzing the robustness of the proposed scheme. It can be observed from Figs.

14(a)-14(d) that the proposed scheme provides better robustness than [12].

(a) (b)

(c) (d)
Fig. 14. (a) Load changes (b) FD of area 1 (c) FD of area 2 (d) TLP exchange of two-area reheated HTPS by load

changes at area 1.

Remark 5: The system parameters of the power system models can be obtained by parameter

identification methods presented in literature such as [36- 38]. However, the system parameters

used in this work are referred from [3, 12] for simplicity.

Table 5
Two areas reheated HTPS for NV

2% load variation in area 1


Methods Area 1 Area 2 Tie-line
( f1 ) ( f 2 ) ( Ptie )
ISE IAE ITAE ISE IAE ITAE ISE IAE ITAE
7.139 7.683 5.755 6.58 2.887 2.074 7.527 1.051 7.503
Proposed method
×10-7 ×10-4 ×10-3 ×10-8 ×10-4 ×10-3 ×10-9 ×10-4 ×10-4
Singh et al. IMC- 7.47 1.35 8.496 2.647 9.069 5.449 2.573 2.54 1.63
PID [12] ×10-5 ×10-2 ×10-2 ×10-5 ×10-3 ×10-2 ×10-6 ×10-3 ×10-2
Singh et al. 5.291 1.243 7.523 1.556 8.309 4.464 1.154 1.809 1.134
Conv. PID [12] ×10-5 ×10-2 ×10-1 ×10-5 ×10-3 ×10-2 ×10-6 ×10-3 ×10-2

Table 6
Two areas reheated HTPS for LB

2% load variation in area 1


Methods Area 1 ( f1 ) Area 2 ( f 2 ) Tie-line ( Ptie )
ISE IAE ITAE ISE IAE ITAE ISE IAE ITAE
7.086 7.655 5.734 5.519 2.541 1.842 5.996 8.813 6.381
Proposed method
×10-7 ×10-4 ×10-3 ×10-8 ×10-4 ×10-3 ×10-9 ×10-5 ×10-4
Singh et al. IMC- 1.039 1.749 1.067 4.395 1.254 7.232 3.804 3.1 2.012
PID [12] ×10-4 ×10-2 ×10-1 ×10-5 ×10-2 ×10-2 ×10-6 ×10-3 ×10-2
Singh et al. 7.503 1.722 9.593 2.871 1.205 6.095 1.868 2.367 1.472
Conv. PID [12] ×10-5 ×10-2 ×10-2 ×10-5 ×10-2 ×10-2 ×10-6 ×10-3 ×10-2

Table 7
Two areas reheated HTPS for UB

2% load variation in area 1


Methods Area 1 ( f1 ) Area 2 ( f 2 ) Tie-line ( Ptie )
ISE IAE ITAE ISE IAE ITAE ISE IAE ITAE
7.307 7.857 5.879 5.78 2.601 1.885 6.159 8.912 6.454
Proposed method
×10-7 ×10-4 ×10-3 ×10-8 ×10-4 ×10-3 ×10-9 ×10-5 ×10-4
Singh et al. IMC- 7.302 1.183 7.621 2.064 7.038 4.383 2.055 2.203 1.43
PID [12] ×10-5 ×10-2 ×10-2 ×10-5 ×10-3 ×10-2 ×10-6 ×10-3 ×10-2
Singh et al. 5.662 1.249 7.452 1.378 6.909 3.818 1.02 1.712 1.069
Conv. PID [12] ×10-5 ×10-2 ×10-2 ×10-5 ×10-3 ×10-2 ×10-6 ×10-3 ×10-2

Table 8
Tie-line power of two area reheated HTPS

TLP exchange (p.u. MW) (Ptie )


Methods 2% load variation in area 1
NV LB UB
Proposed method -1.244×10-8 -1.605×10-8 -7.615×10-9
Singh et al. IMC-PID [12] 2.54×10-5 5.038×10-6 1.002×10-5
Singh et al. IMC-PID [12] -4.483×10-6 -1.828×10-5 -1.38×10-5
4.5. Three-area reheated HTPS extension with wind and solar energy penetrations

The proposed method is extended to three-area multi-source power system, which consists of

reheated HTPS [12] with wind and solar energy penetrations by considering random wind speed

and random irradiance change [39, 40] as shown in Fig. 15(a)-15(d). The predictive model,

DRC, OLC and the values of parameters of three-area reheated HTPS, are referred from the two-

area reheated HTPS as given in Section 4.3. A wind power generator with random speed [40] is

considered as shown in Fig. 15(c). The mechanical torque of wind generator is developed by

wind energy which transfers through shaft to rotor of generator. The net output power of

generator is developed by random wind speed which is used to reheat the HTPS. The WPG

mathematical equations are referred from [39]. The solar power penetration model is designed as

shown in Fig. 15 (d). The fluctuation in solar power is obtained by using solar irradiance model

which uses a white noise block.

Reheated HTPS+Wind and Reheated HTPS+Wind and


Solar penetrations Solar penetrations

AREA-2 AREA-3

AERA-1

Reheated HTPS+Wind and


Solar penetrations

Fig. 15 (a) Simplified model of three-area power system


Pwi Ppvi
Wind power Solar power
+ +
Reheated thermal power

1 (1  sK riTri ) 1 Pdi
(1  sTgi ) (1  sTri ) (1  sTti ) _
ui + K pi
+ Ggi Gri Gti + fi
_ + _ (1  sTpi )
+
1 (1  sTHRi ) (1  sTwi )
G pi
_ (1  sTHgi ) (1  sTHi ) (1  0.5sTwi )
1 GHgi GHRi GHwi
Ri 1
Rj Hydro power

Bi
+
ACEi Tij
PTie s _

f j i
Fig. 15 (b) Block diagram of ith -area power system

+ 1
Wind speed
_ s

Saturation Saturation
_ + K Pitch angle Wind Torque
Constant
_ s turbine
+

+
Generator speed K1
s _
Saturation Generator model

_
+
 + K2 
_ s
Power

Lookup table Output power

Fig. 15 (c) Wind power penetration model


1
12 s  1
 +  K spg Solar power

+ Tspg s  1
White-noise Low-pass filter 

0.6*sqrt

Psolar initial

Fig. 15 (d) Solar power penetration model

The random load disturbance, random wind speed and solar irradiance are applied to the

model, which are plotted in Fig. 16(a)-16(c). The random load disturbance of the amplitudes 0.1,

-0.1, 0.15 and 0.05 are applied at t = 10, 30, 50, and 70 seconds to area 1 for analyzing the

robustness of the proposed scheme.

In Section 4.4, the proposed method is implemented for LFC of two area reheated HTPS

without any external renewable power sources and better results are achieved with fewer

oscillations. Therefore, in this section, the proposed method is applied for LFC of three-area

reheated HTPS with WPG and SPG power along with random wind speed and solar irradiance.

The simulation results of proposed method for three-area reheated HTPS along with random

wind speed and solar power penetration is shown during the random load disturbance for two

cases: (i) with proposed controller; and (ii) without proposed controller. It is observed from Figs.

16(a)-16(f) that the proposed method is capable of reducing the frequency and tie-line power

deviations with minimum oscillations significantly and gives a satisfactory load disturbance

rejection as compared to the without controller case.


(a)

(b) (c)
Fig. 15 Plots of (a) random load disturbance (b) wind speed and power, (c) Solar irradiance and power
(a) (b)

(c) (d)
(e) (f)
Fig. 16 Responses of FD and TLP of three-area multi-unit multi-sources reheated HTPS with WPG and SPG (a)
Area-1 (b) Area-2 (c) Area-3 (d) Area-4, and (e)- (h) Tie-line power for three area.

4.6. Four-area reheated HTPS extension with renewable power generations

The proposed method is extended to four-area multi-source power system, which consists of

reheated HTPS [12] with renewable power generations (RPGs) such as wind power generation

(WPG), solar power generation (SPG), fuel cell (FC) and aqua-electrolyzer (AE) as shown in

Fig. 17(a)-17(b). The models of renewable power sources are referred from [41-44] which is a

series combination of linearized models of WPG ( Gwpg ), SPG ( Gspg ), FC ( G fc ) and AE ( Gae ).

The values of parameters of system are given as below:

K p1  K P 2  K p 3  K P 4  120, Tp1  Tp 2  Tp 3  Tp 4  20, Tt1  Tt 2  Tt 3  Tt 4  0.3,


Tg1  Tg 2  Tg 3  Tg 4  0.08, Tr1  Tr 2  Tr 3  Tr 4  10, K r1  K r 2  K r 3  K r 4  0.5,
Tw1  Tw 2  Tw3  Tw 4  1, THg1  THg 2  THg 3  THg 4  41.6, THR1  THR 2  THR 3  THR 4  5,
TH 1  TH 2  TH 3  TH 4  0.513, R1  R2  R3  R4  R5  R6  R7  R8  2.4,
B1  B2  B3  B4  0.425, T12  T23  T34  T41  0.545, Pr1  Pr 2  Pr 3  Pr 3  2000.
K wpg1  K wpg 2  K wpg 3  K wpg 4  1, K spg1  K spg 2  K spg 3  K spg 4  1,
K fc1  K fc 2  K fc 3  K fc 4  1/100, K ae1  K ae 2  K ae3  K ae 4  1/ 500,
Twpg1  Twpg 2  Twpg 3  Twpg 4  1.5, Tspg1  Tspg 2  Tspg 3  Tspg 4  1.8,
T fc1  T fc 2  T fc 3  T fc 4  4, Tae1  Tae 2  Tae3  Tae 4  0.5;
Reheated HTPS+RPGS Reheated HTPS+RPGS

AREA-2 AREA-3

AERA-1 AREA-4

Reheated HTPS+RPGS Reheated HTPS+RPGS

Fig. 17 (a) Simplified model of four-area power system

Fig. 17 (b) Block diagram of four-area power system


The transfer function of area 1 of four area power with respect to disturbance input is obtained as

2.55s10  59.62 s 9  467.42 s8  1777.78s 7  3703.71s 6  4372.25s 5


2880.73s 4  1010.67 s 3  169.93s 2  11.212 s  0.1844
G4 ( s )  11 .
s  23.427 s10  184.495s 9  706.215s8  1487.132 s 7  1818.582 s 6
1215.317 s 5  452.731s 4  86.441s 3  7.727 s 2  0.2921s  0.00363
From Section 3.1, the predictive model is obtained as

N (s) 2.55s 2  39.099 s  1.2644


P4 m ( s )   3 .
D( s ) s  2.701s 2  1.258s  0.02489

The proposed controllers for LFC of four-area power system are obtained as

( s 3  2.701s 2  1.258s  0.02489)(0.0002 s 3  0.0096 s 2  0.16 s  1)


QD ( s )  and
(2.55s 2  39.099 s  1.2644)(0.04 s  1) 4

0.0002 s 6  0.01014 s 5  0.1862 s 4  1.444 s 3  2.903s 2  1.262 s  0.02489


GC ( s )  .
0.000006528s 6  0.0002427 s 5  0.002189 s 4  7.81e 5 s 3  2.193e 18 s 2

The proposed controller is implemented in each area of the four-area power system and 1% ELD

is applied in area 1 at t  1 sec. The simulation results are shown in Figs. 18(a)-18(d) for

frequency deviation and Figs. 18(e)-18(h) for tie-line power. It is observed from the figures that

the proposed scheme provides significantly better results than the conventional PID controller (

GC ( s )  9.9  15.9 / s  3.0 s. ) which is tuned by using Genetic algorithm [45].

(a) (b)
(c) (d)

(e) (f)

(g) (h)

Fig. 18 Responses of FD and TLP of four-area multi-unit multi-sources reheated HTPS with RPGs (a) Area-1 (b)
Area-2 (c) Area-3 (d) Area-4, and (e)- (h) Tie-line power for four area.
5. Conclusion

This article presents a novel dual loop-internal model control approach for LFC of single and

multi-area power systems. The proposed scheme utilizes a predictive model which is obtained by

model reduction and conventional IMC scheme. In order to establish the superiority of the

proposed scheme, the results are compared with existing techniques. Further, the robustness of

the proposed scheme is verified by inserting 50% parametric uncertainty and applying different

combinations of external load disturbances. The simulation results of a three-area reheated

hydro-thermal power system model with wind and solar energy penetrations and a four area

reheated hydro-thermal power system with renewable power sources such as wind power

generator, solar power generator, fuel cell and aqua-electrolyzer are also shown. It is observed

from the results that the proposed scheme gives a better disturbance rejection and fewer

oscillations. The proposed scheme can also be utilized for process control and biomedical

engineering applications.

References

[1] Kundur P. Power System Stability and Control. New York NY USA: Tata McGraw-Hill, 2012.
[2] Tan W. Tuning of PID load frequency controller for power systems. Energy Convers Manag
2009;50(6):1465–1472.
[3] Tan W. Unified tuning of PID load frequency controller for power systems via IMC. IEEE Tran Pow Sys
2010;25(1):341–350.
[4] Tan W. Load frequency control: problems and solutions. In: Proceedings of the 30th Chinese control
conference Yantai China 2011;6281–6286.
[5] Saxena S, Hote YV. Load frequency control in power systems via internal model control scheme and model-
order reduction. IEEE Tran Pow Sys 2013;28(3):2749–2757.
[6] Saxena S, Hote YV. Advances in Internal Model Control Technique: A Review and Future Prospects. IETE
Tech Review 2012;29(6):461–472.
[7] Saxena S, Hote YV. Decentralized PID load frequency control for perturbed multi area power systems. Electr
Power Energy Syst 2016;81:405–415.
[8] Liu T, Gao F. New insight into internal model control filter design for load disturbance rejection. IET Control
Theory Appl 2010;4(3):448–460.
[9] Xiao L, Gang C, Wang YG. IMC-PID Controller Design for Power Control Loop Based on Closed-loop
Identification in the Frequency Domain. In: Proceedings of the 14th IFAC Symposium on Large Scale
Complex Systems: Theory and Applications Riverside California USA IFAC 2016;49(4):79–84.
[10] Ghousiya BK, Seshagiri RA, Radhakrishnan TK. Enhanced IMC based PID controller design for non-
minimum phase (NMP) integrating processes with time delays. ISA Tran 2017;68:223–234.
[11] Sonker B, Kumar D, Samuel P. A modified two-degree of freedom-internal model control configuration for
load frequency control of a single area power system. Turk J Elec Eng Comp Sci 2017;25:4624 – 4635.
[12] Singh J, Chattterjee K, Vishwakarma CB. Two degree of freedom internal model control-PID design for LFC
of power systems via logarithmic approximations. ISA Tran 2018;72:185–196.
[13] Anwar N, Pan S. A new PID load frequency controller design method in frequency domain through direct
synthesis approach. Electr Power Energy Syst 2015;67:560–569.
[14] Shabani H, Vahidi B, Ebrahimpour M. A robust PID controller based on imperialist competitive algorithm for
load-frequency control of power systems. ISA Tran 2013;52:88–95.
[15] Padhan DG, Majhi S. A new control scheme for PID load frequency controller of single-area and multi-area
power systems. ISA Tran 2013;52:242–251.
[16] Sathya MR, Ansari MMT. Load frequency control using BAT inspired algorithm based dual mode gain
scheduled of PI controllers for interconnected power system. Electr Power Energy Syst 2015;64:365–374.
[17] Dash P, Saikia LC, Sinha N. Automatic generation control of multi area thermal system using Bat algorithm
optimized PD-PID cascade controller. Electr Power Energy Syst 2015;68:364–372.
[18] Rahman A, Saikia LC, Sinha N. Load frequency control of thermal system under deregulated environment
using biogeography based optimized three degree of freedom integral derivative controller. IET Gener
Transm Distrib 2015;9(15):2284–2293.
[19] Rahman A, Saikia LC, Sinha N. Automatic generation control of an unequal four area thermal system using
biogeography based optimized 3DOF-PID controller. IET Gener Transm Distrib 2016;10(16):1–31.
[20] Raju M, Saikia LC, Sinha N. Automatic generation control of a multi-area system using ant Lion optimizer
algorithm based PID plus second order derivative controller. Electr Power Energy Syst 2016;80:52–63.
[21] Sahu RK, Panda S, Rout UK, Sahoo DK. Teaching learning based optimization algorithm for automatic
generation control of power system using 2-DOF PID controller. Electr Power Energy Syst 2016;77:287–301.
[22] Sahu RK, Panda S, Biswal A, Sekhar GTC. Design and analysis of tilt integral derivative controller with filter
for load frequency control of multi-area interconnected power systems. ISA Tran 2016;16:251–264.
[23] Sondhi S, Hote YV. Fractional order PID controller for load frequency control. Energy Convers Manag 2014;
85:343–353.
[24] Zamani A, Barakati SM, Darmian SY. Design of a fractional order PID controller using GBMO algorithm for
load–frequency control with governor saturation consideration. ISA Tran 2016;64:56–66.
[25] Zheng Y, Zhou J, Xu Y, Zhang Y, Qian Z. A distributed model predictive control based load frequency
control scheme for multi-area interconnected power system using discrete-time Laguerre functions. ISA Tran
2017; 68:127–140.
[26] Mi Y, Fu Y, Li D, Wang C, Loh PC, Wang P. The sliding mode load frequency control for hybrid power
system based on disturbance observer. Int J of Electrical Power & Energy Systems 2016; 74: 446-452.
[27] Cui Y Xu L, Fei M, Shen Y. Observer based robust integral sliding mode load frequency control for wind
power systems. Control Engineering Practice 2017; 65: 1-10.
[28] Prakash S, Sinha SK. Load frequency control of multi-area power systems using Neuro-fuzzy hybrid
intelligent controllers. IETE J of Research 2015; 61: 526-532.
[29] Sivalingam R, Chinnamuthu S, Dash SS. A modified whale optimization algorithm-based adaptive fuzzy
logic PID controller for load frequency control of autonomous power generation systems. Automatika
2017;58:410-421.
[30] Arya Y, Kumar N. Design and analysis of BFOA-optimized fuzzy PI/PID controller for AGC of multi-area
traditional/restructured electrical power systems. Soft Computing 2017; 21: 6435–6452.
[31] Arya Y. Automatic generation control of two-area electrical power systems via optimal fuzzy classical
controller. Journal of the Franklin Inst 2018;355:2662–2688.
[32] Prakash S, Sinha SK. ALFC of hybrid multi-generation power system using UC and TCPS by ANFIS control
technique. Int J of Electro 2018:1-38.
[33] Prakash S, Sinha SK. Automatic load frequency control of six areas’ hybrid multi-generation power systems
using Neuro-fuzzy intelligent controller. IETE J of Research 2018;64: 471-481.
[34] Sonker B, Kumar D, Samuel P. Design of two degree of freedom-internal model control configuration for
load frequency control using model approximation. Int J Mod Sim 2018;1–11.
[35] Singh J, Vishwakarma CB, Chattterjee K. Biased reduction method by combining improved modified pole
clustering and improved Padé approximations. Appl Math Model 2016;40:1418–1426.
[36] Kishor N and Singh S. Simulated response of NN based identification and predictive control of hydro plant.
Expert Systems with Applications. 2007; 32: 233–244.
[37] Smith J, Fatehi F, Woods C. Transfer function identification in power system applications. IEEE Trans on
Power Systems. 1993; 8(3): 1282–1290.
[38] Li C, Zhou J, Li Q, An X, Xiang X. A new T–S fuzzy modelling approach to identify a boiler–turbine system.
Expert Systems with Applications. 2010; 37(3): 2214–2221.
[39] Liu Y, Wu QH, Zhou XX. Co-ordinated multi loop switching control of DFIG for resilience enhancement of
wind power penetrated power systems. IEEE Trans Sustain Energy 2016;7:1089–1099.
[40] Magdy G, Mohamed EA, Shabib G, et al. SMES based a new PID controller for frequency stability of a real
hybrid power system considering high wind power penetration. IET Renew Power Gener 2018;12:1304–1313.
[41] Dulal CD, Roy AK., Sinha N. GA based frequency controller for solar thermal–diesel–wind hybrid energy
generation/energy storage system. Electrical power and energy systems 2012;43:262–279.
[42] Shankar G, Mukherjee V. Load frequency control of an autonomous hybrid power system by quasi-
oppositional harmony search algorithm. Electrical power and energy systems 2016;78:715–734.
[43] Lee D, Wang L. Small-Signal Stability Analysis of an Autonomous Hybrid Renewable Energy Power
Generation/Energy Storage System Part I: Time-Domain Simulations. IEEE Transactions on Energy
Conversion 2008; 23(1): 311-320.
[44] Sanki P, Babu M. New approach in two-area interconnected AGC including various renewable energy sources
using PSO. Turk J Elec Eng & Comp Sci 2018; 26:1491-1504.
[45] Goldberg DE. Genetic algorithms. Pearson Education India, 2006.

You might also like