Adaptive PI Controller for Voltage Regulation in Power Systems

Research article
Adaptive PI controller to voltage regulation in power systems:
STATCOM
as a case study
Mohammad Reza Tavana, Mohammad-Hassan Khooban n
, Taher Niknam
Department of Electrical Engineering, Shiraz University of Technology, Iran
a r t i c l e i n f o
Article history:
Received 1 January 2016
Received in revised form
18 August 2016
Accepted 29 September 2016
Available online 25 November 2016
Keywords:
Online intelligent control
General Type-2 Fuzzy Logic Control
Static synchronous compensator (STAT-
COM)
Teaching Learning Based Optimization
(TLBO)
Voltage regulation
a b s t r a c t
Static synchronous compensator (STATCOM) provides the means to improve quality and reliability of a
power system as it has the functional capability to handle dynamic disturbances, such as transient
stability and power oscillation damping as well as to providing voltage regulation. In this paper, a robust
adaptive PI-based optimal fuzzy control strategy is proposed to control a STATCOM used in distribution
systems. The proposed intelligent strategy is based on a combination of a new General Type-II Fuzzy
Logic (GT2FL) with a simple heuristic algorithm named Teaching Learning Based Optimization (TLBO)
Algorithm. The proposed framework optimally tunes parameters of a Proportional-Integral (PI) controller
which, similar to most of other researchers regarding control of STATCOM, are in charge of controlling
the device. The proposed controller guaranties robustness and stability against uncertainties caused by
external disturbances or ever-changing nature of the power systems. The TLBO optimizes the parameters
of the controller as well as the input and output membership functions. To validate the efficiency of the
proposed controller, the obtained simulation results are compared with those of the two most recent
researches applied in this field, namely, conventional Proportional Integral (PI) controller and Optimal
Fuzzy PI (OFPI) controller. Results demonstrate the successfulness and effectiveness of the proposed
online-TLBO General Type-2 Fuzzy PI (OGT2FPI) controller and its superiority over conventional ap-
proaches.
& 2016 ISA. Published by Elsevier Ltd. All rights reserved.
1. Introduction
Over the recent decades, power systems are experiencing rapid
changes caused by deregulation of the power industry as well as
fast growing demand. As a result, new considerations should be
taken into account to ensure stability and reliability of the power
systems [1]. Accordingly, new apparatuses are introduced to
handle such problems and, consequently, improve the perfor-
mance of the power grids. Flexible Alternative Current Transmis-
sion Systems (FACTS) devices are one of the possible solutions to
the above-mentioned issues. FACTS devices are able to enhance
the controllability of both active and reactive power. If used
alongside with proper controllers, FACTS devices will be able to
improve the dynamic reaction of the system while facing transient,
small signal and voltage stability [1].
Static synchronous compensator (STATCOM) is a shunt FACTS
device that takes advantage of power electronics to control the
flow of power to enhance transient stability of power grids [2]. To
regulate voltage at its terminals, a STATCOM device controls the
volume of reactive power that is injected into or absorbed from
the grid. When the voltage of the grid is lower than desired, the
STATCOM generates reactive power (capacitive STATCOM) to in-
crease the voltage. Inversely, when system voltage is higher than
expected, it absorbs proper amount of reactive power (inductive
STATCOM) to decrease the voltage.
So far, many different control methods have been proposed to
control STATCOM dynamic [3–11]. The majority of the presented
approaches in the literature include some sort of linearization of
equations of the voltage source converters (VSC) average value
model at a specific equilibrium [3–11]. A STATCOM based on pro-
portional-integral (PI) controller was first proposed in [3]. Since
then, several different PI controllers have been presented in the
literature. Results reported in the literature indicate that satisfac-
tory performance will be obtained from PI based STATCOM if the
parameters are tuned properly, no matter if a two-level converter
[4–6] is used or a multilevel converter is applied [7–11]. The main
drawback of PI based STATCOM is that its performance might
decline gravely if operating conditions changes from the assumed
Contents lists available at ScienceDirect
journal homepage: www.elsevier.com/locate/isatrans
ISA Transactions
http://dx.doi.org/10.1016/j.isatra.2016.09.027
0019-0578/& 2016 ISA. Published by Elsevier Ltd. All rights reserved.
n
Corresponding author.
E-mail addresses: khooban@sutech.ac.ir,
mhkhoban@gmail.com (M.-H. Khooban).
ISA Transactions 66 (2017) 325–334

situations, especially in the case of a large disturbance like sudden
load variation or a short-circuit fault nearby.
So far, a few researches have addressed assigning gains of the
STATCOM PI controller to enhance voltage stability and to decrease
the time required to tune the gains. In [3,6,12–16], linear quadratic
regular (LQR) based linear optimal controls are presented. Such
approaches highly rely on the expertise of the designer to assign
optimal parameters. On the other hand, in [4], an STATCOM state
feedback framework is presented which is based on a zero set
concept. Again, similar to pervious works such as [12–14], gains of
the state feedback controller are heavily relied on the designer's
choice and experience. In [14–16], a fuzzy based PI controller is
introduced to properly tune the gains of the PI controller. How-
ever, the proposed approach still depends on designer's initial
choices of the actual and deterministic gains. In [17], a population-
based optimization algorithm is presented to optimally choose and
tuning the gains of the PI controller. However, the method will
have difficulties for being applied in the real-world applications as
it usually takes a long time to calculate the gains of the controller.
It seems that a tradeoff between performance and the variety of
possible operation conditions still should be struck during the
process of designer's decision-making. Therefore, efficient perfor-
mance may not be always obtainable in a certain operating
condition.
Fuzzy Logic is widely applied in different applications such as
system identification, control and modeling of nonlinear dynamic
systems [18,19]. In researches such as [20,21], several controllers
based on combination of fuzzy logic with PI or PID controllers are
presented and different approaches are proposed to improve the
performance of the resulted hybrid controller. It should be noted
that tuning coefficients of PID controllers might be a difficult, time
consuming and costly procedure [22,23]. In fact, a proficient gainer
regularly tries to control the process by adjusting different gains of
controller according to output error and/or rate of error.
Lately, a large number of researchers pay attention to general
type-2 fuzzy sets and systems because they have the capacity to
deal with uncertainties [24–27]. The first one who presented Type-
2 fuzzy sets as a development of type-1 fuzzy sets was Zadeh in
1975 [26]. Since then, type-2 fuzzy logic systems especially IT2FLSs
due to their calculation simplicity have been successfully applied
to engineering areas. This illustrates that when IT2FLSs face with
different uncertainties such as dynamic uncertainties, rule un-
certainties, external disturbances and noises, they do better in
comparison with type-1 fuzzy logic systems (T1FLS). See [27]. The
rules which are made in a fuzzy logic system by the use of ac-
cessible information may be uncertain. Unlike interval type-2
fuzzy sets (IT2FS) and type-1 fuzzy sets (T1FS), General type-2
fuzzy sets can deal with rule uncertainties usefully. As general
type-2 fuzzy sets and systems are computationally complex, just
IT2FLSs have been mainly used in literatures so far. But, Liu [28]
proposed a useful way to compute centroid and type reduction of
GT2FLS by using a newly introduced plane representation
theorem.
In this paper an online TLBO based General Type 2 Fuzzy
Tuning PI (OGT2FPI) controller is presented. The proposed fra-
mework is simple and does not have complexities of the pre-
viously presented methods in the literature. The parameters of
input and output membership functions of the fuzzy controller's
coefficients are optimized simultaneously by TLBO. The proposed
framework can be easily utilized in PI controllers of STATCOM in
power systems. Performance of the proposed controller is eval-
uated on a test case power system owning a ±100-MVAR STAT-
COM with a 48-pulse VSC and connected to a 500-kV bus. Simu-
lation results demonstrate the superiority of the proposed con-
troller compared to OFPI and conventional PI controller.
2. The dynamic model of STATCOM
2.1. Advantages of STATCOM in power systems
With the ongoing deregulation of the electric utility industry,
numerous changes are continuously being introduced to a once
predictable business. With electricity increasingly being con-
sidered as a commodity, transmission systems are being pushed
closer to their stability and thermal limits while the focus on the
quality of power delivered is greater than ever. In addition, dy-
namic reactive power support is becoming more important,
especially in urban areas where local (i.e., at the load) generation is
being reduced or eliminated. In the deregulated utility environ-
ment, financial and market forces will demand a more optimal and
profitable operation of the power system with respect to genera-
tion, transmission, and distribution. Now, more than ever, ad-
vanced technologies are paramount for the reliable and secure
operation of power systems. Power electronic based equipment,
such as FACTS controllers, with their capability to rapidly respond
to system events, increase power transfer limits, and improve the
quality of power delivered, constitute one of the most-promising
technical advancements to address the new operating challenges
being presented today [1,2]. This paper focuses on one such FACTS
controller for voltage support as a dynamic reactive power (var)
source, namely the static compensator (STATCOM). Typically, the
first step in providing voltage/var support in a power system is
with the application of shunt capacitors, shunt reactors, and
transformer on-load-taps (OLT). Therefore, combined control of
this discrete-acting existing equipment with a continuously con-
trolled dynamic device, such as a STATCOM, becomes important
for the overall technical and economic operation of the system
[10].
2.2. STATCOM description
Fig. 1 demonstrates the equivalent circuit of a STATCOM. In this
scheme, the series resistance Rs represents the sum of the inverter
conduction losses and the transformer winding resistance losses.
The leakage inductance of the transformer is represented by Ls.
The resistance Rc which is in shunt with the capacitor C also de-
note for the losses related to the switching of the inverter as well
as the losses of the capacitor. Vas, Vbs and Vcs also denote the three-
phase output voltages of STATCOM; moreover, Val, Vbl and Vcl re-
present voltages of the three- phase bus; and, ias, ibs and ics are the
output currents of the three-phase STATCOM [2,6].
2.3. The dynamic model of STATCOM
Mathematical model of a three-phase STATCOM can be given as
follows [2,4,6]:
= − + −
( )
L
di
dt
R i V V
1s
as
s as as al
= − + −
( )
L
di
dt
R i V V
2s
bs
s bs bs bl
= − + −
( )
L
di
dt
R i V V
3s
cs
s cs cs cl
( )
( )= − + + −
( )
⎛
⎝
⎜
⎞
⎠
⎟ ⎡⎣ ⎤⎦
d
dt
CV t V i V i V i
V t
R
1
2 4
dc as as bs bs cs cs
dc
c
2
2
Taking advantage of abc/dq transformation, Eqs. (1) to (4) can
be simplified as follows:
M.R. Tavana et al. / ISA Transactions 66 (2017) 325–334326

ω α
ω α
α α
=
−
− −
− − −
−
( )
⎡
⎣
⎢
⎢
⎢
⎤
⎦
⎥
⎥
⎥
⎡
⎣
⎢
⎢
⎢
⎢
⎢
⎢
⎢
⎤
⎦
⎥
⎥
⎥
⎥
⎥
⎥
⎥
⎡
⎣
⎢
⎢
⎢
⎤
⎦
⎥
⎥
⎥
⎡
⎣
⎢
⎢
⎢
⎤
⎦
⎥
⎥
⎥
d
dt
i
i
V
R
L
K
L
R
L
K
L
K
C
K
C R C
i
i
V
L
V
V
cos
sin
3
2
cos
3
2
sin
1
1
0
5
ds
qs
ds
s
s s
s
s s
c
ds
qs
ds
s
dl
ql
where ids and iqs respectively represent the currents of d and q
corresponding to ibs, ibs, and ics; the relation between the dc vol-
tage and the peak phase-to-neutral voltage on the ac side is de-
noted by factor K; Vds represents the voltage of the dc-side; α
denoted the phase angle at which the voltage of the STATCOM
leads the bus voltage; speed of the rotating angle of the voltage
vector is denoted by ω; and Vdl and Vql are respectively the d and q
axis voltage corresponding to Val,Vbl and Vcl. Since Vql is assumed
to be 0, considering the definition of the instantaneous active and
reactive power, (6) and (7) can be obtained as bellow [2,4]:
=
( )
dl V i
3
2 6dl ds
=
( )
ql V i
3
2 7dl qs
Given to the above mentioned equations, the traditional control
strategy can be obtained, and accordingly, the control block dia-
gram of the STATCOM is shown in Fig. 2 [2].
As demonstrated by Fig. 2, the phase-locked loop (PLL) is in
charge of providing a synchronizing signal that will be used as the
reference angle for the measurement system. Measured voltage of
the bus line Vm is compared to the reference voltage Vref, and
accordingly reactive reference current Iref will be provided by the
voltage regulator. The allowable voltage error at the rated reactive
current flow is denoted by droop factor Kd. The reactive current of
the STATCOM Iq is compared with reference current Iqref, and the
output of the current regulator is calculated using the the angle
phase shift between the inverter voltage and the the system vol-
tage. Given the upper bound for reactive power that can be gen-
erated by the STATCOM, a limit will be imposed on the value of
control that is denoted by the value of limiter.
3. General Type-II Fuzzy Logic Control
A GT2FS in a universal set X can be defined as
∫μ̃=
( )
∈ ( )
̃
A
x x
x X
/
8
A
∫μ ( )=
( )
∈
∈[ ]
( )
̃ x
f u
u
u J
J, 0, 1
9
A
x
x
x
where in this formula μ ( )̃ xA is called a secondary membership
function (MF) and ( )f ux is called secondary grade; Jx is the domain
of the secondary MF which is called primary membership and u is
a fuzzy set in [0, 1]. Fig. 3 illustrates a GT2FS where the upper and
lower MFs are triangular and its secondary MF is also triangular.
When ( )=f u 1x IT2FS is obtained that demonstrate a uniform un-
certainty in the primary membership function and is simply
Fig. 1. The general scheme of STATCOM.
Fig. 2. The block diagram of PI controller for STATCOM.
M.R. Tavana et al. / ISA Transactions 66 (2017) 325–334 327

described by its lower μ̲ ( )˜ xA
and upper μ¯ ( )˜ xA MFs. Because of
calculation simplicity, especially in the type reduction, many re-
searchers use interval type-2 fuzzy sets instead of general type-2
fuzzy sets [25,28,29] (Fig. 3).
Lately, Liu [28] presented a new method for GT2FSs which is
theoretically and computationally effective. Because this method
resembles the α-cut for type-1 fuzzy sets, it is named a α-plane for
type-2 fuzzy sets. ̃αA is the denotation of An α-plane representa-
tion for a GT2FS ̃A. It is the union of all primary MFs whose sec-
ondary grades are greater than or equal to the special value α:
∫μ̃ =
( )
∈ ( )
α
̃αA
x x
x X
/
10
A
∫μ
α
( )=
( )≥
∈
∈[ ]
( )
̃α
x
f u
u
u J
J, 0,1
11
A
x
x
x
Then a GT2FS ̃A based on α-plane representation theorem can
be demonstrated in the following form:
α̃= ⋃ ̃
( )α
α
∈[ ]
A A/
120,1
It is a beneficial representation because α ̃αA/ can be seen as an
IT2FS with the secondary grade of level α. As a result, several
IT2FSs may be made from the decomposition of a general type-2
fuzzy set with a corresponding level of α for each, where
α = { … ( − ) }K K K0,1/ , , 1 / , 1 . In simpler terms, a general type-2
fuzzy logic system can be seen as a huge collection of IT2FLSs with
one IT2FLS for each value of α. However, Liu [30] showed that
using only 5 to 10 α-plane can get the required accuracy for cen-
troid calculation. Fig. 4 illustrates the new designing for a general
type-2 fuzzy system based on α-plane representation.
In general, a GT2FLS is made of a fuzzifier; fuzzy rule-based;
fuzzy inference engine; type reducer and defuzzifier. Fuzzifier
maps real values into fuzzy sets. Singleton fuzzifier whose output
is a single point of a unity membership grade is used in this paper
because it is simple. Fuzzy rule base includes fuzzy IF-THEN rules.
In the following The jth rule in the GT2FLS is shown:
̃ ̃ … ̃ ̃
= … ( )
R If x is F and x is F and x is F Then y is G
j M
:
1, 2, 13
j j j
n n
j j
1 1 2 2
where xi for = …i n1, , and y are the input and output of the
GT2FLS, ̃Fi
j
and ̃G
j
are general type-2 antecedent and the con-
sequent sets. A mapping from input GT2FSs to output GT2FSs is
given by the inference engine that merges rules. Because α-plane
representation for fuzzy set is used, the firing set for each related
IT2FS is shown as following:
( ) ( ) ( )= ̲ ¯
( )α α α
⎡
⎣⎢
⎤
⎦⎥F X f X f X,
14
j j j
( ) μ μ μ̲ = ̲ * ̲ *…* ̲
( )α ˜ ˜ ˜
α α α
f X
15
j
F F F
j j
n
j
1 2
( ) μ μ μ¯ = ¯ * ¯ *…* ¯
( )α ˜ ˜ ˜
α α α
f X
16
j
F F F
j j
n
j
1 2
Here ( )̲ α
f Xj
and ( )¯
αf X
j
are the lower and upper MFs of the jth
rule with level of α, and * indicates product t-norm. a type reducer
changes the output of the inference engine which is a type-2 fuzzy
-3 -2 -1 0 1 2 3
0
0.2
0.4
0.6
0.8
1
x
(x)
0 0.2 0.4 0.6 0.8 1
0
0.2
0.4
0.6
0.8
1
(x)
f()
Fig. 3. a general type-2 fuzzy set with triangular upper and lower MFs where the secondary MF is triangular.
Fig. 4. Architecture for a general type-2 fuzzy Logic system.

set into into a type-1 fuzzy set before defuzzification. Five kinds of
reducers which are based on calculating the centroid of an IT2FS
are demonstrated in [31]. The output of the type reduction in
IT2FLS is defined with its left-end point yl and right-end point yr
due to uniformly secondary grade of IT2FLS.
KM iterative algorithms, introduced two algorithms for calcu-
lating these two end points in [32], is presented by Mendel and
Karnik. In comparison to the other type reduction methods, center
of sets (COS) is used a lot because of its computation simplicity by
the KM iterative algorithm [33]. If singleton fuzzifier is used,
product inference engine and COS type reducer, left and right end
points for each part of GT2FLS based on α- representation theorem
can be shown as follows (Fig. 5).
θ
θ ξ=
∑ ¯ + ∑ = +
∑ ¯ + ∑ ̲
=
( )
α
α α
= = +
= = +
α α αy
f j L
f f
1
17
l
j
L j
l
j
j L
M
j
L j
j L
M j
l
T
l
1 1
1 1
a
where in this formula θ αl
j
is the left-end point of jth consequent set
with level of α, θ θ θ= …α α α
⎡
⎣
⎤
⎦, ,l l l
M T1
, ξ =
̲ ¯
α
α
α
α
α
⎡
⎣
⎢
⎤
⎦
⎥,l
j f
D
f
D
j
l
j
l
,
= ∑ ¯ + ∑ ̲α α= = +αD f fl j
L j
j L
M j
1 1
and ξ ξ ξ= …α α α
⎡
⎣
⎤
⎦, ,l l l
M T1
. In addition,
θ θ
θ ξ=
∑ ¯ + ∑ ̲
∑ ¯ + ∑ ̲
=
( )
α α
α α
= = +
= = +
α α αy
f f
f f 18
r
j
R j
r
j
j R
M j
r
j
j
R j
j R
M j
r
T
r
1 1
1 1
a a
where θ αr
j
is the right end point of jth consequent set with level of
α, θ θ θ= …α α α
⎡
⎣
⎤
⎦, ,r r r
M T1
, ξ =
̲ ¯
α
α
α
α
α
⎡
⎣
⎢
⎤
⎦
⎥,r
j f
D
f
D
j
r
j
r
, = ∑ ¯ + ∑ ̲α α= = +αD f fr j
R j
j R
M j
1 1
and ξ ξ ξ= …α α α
⎡
⎣
⎤
⎦, ,r l l
M T1
. Meanwhile, performing KM iterative al-
gorithm can specify R and L for each individual IT2FLS of level α.
From the combination of all of these obtained intervals into a type-
1 fuzzy set like Fig. 5, a crisp output can be obtained using centroid
defuzzification as:
∑
α
α
α α= =
∑ ( + )
∑
= ( + ) +
= …
−
( )
α
α
α α
α α
⎪ ⎪
⎪ ⎪
⎧
⎨
⎩
⎫
⎬
⎭
U y
y y
y y K
K
K
K
2
/ 1,
0.
1
, ,
1
, 1
19
Fuzzy
l r
a
l r
where Kþ1 shows the number of the α-planes or in other
words it determines the number of individual IT2FLSs.
4. Teaching Learning Based Optimization (TLBO)
TLBO algorithm, originally developed by Rao et al. [34,35], is a
population-based optimization algorithm. In TLBO algorithm a
population of solutions is utilized to proceed to the global solution.
The two elementary components of this algorithm are Teacher and
Learners. Based on two basic modes of the learning, through tea-
cher (known as teacher phase) and interacting with the other
learners (known as learner phase), the procedure of TLBO is di-
vided into two parts, 1) Teacher phase and 2) Learner phase.
4.1. Teacher phase
The first part of the algorithm is teacher phase where learners
learn through the teacher. During this phase, a teacher endeavors
to increase the mean result of the class room from any value. Since
it is practically impossible, a teacher can improve the mean of the
class room to other better value depending on the class capability.
Consider Mi be the mean and Ti be the teacher at any iterationi. Ti
will try to enhance existing mean Mitowards it so the new mean
will be Ti designated as Mnew. The solution is updated according to
the difference between the existing and the new mean given by
[34]
_ = ( − ) ( )Difference Mean r Mean T M 20i i new F i
where TF is the teaching factor which decides the value of mean to
be changed, and ri is the random number in the range [0, 1]. The
teaching factor TF is generated randomly during the algorithm in
the range of 1–2, in which 1 corresponds to no increase in the
knowledge level and 2 corresponds to complete transfer of
knowledge. The value of TF is randomly opted with equal prob-
ability as
( ){ }= [ + − ] ( )T round rand1 0, 1 2 1 21F
According to this _Difference Mean, the existing solution is up-
dated as follows:
= + _ ( )X X Difference Mean 22new i old i j, ,
4.2. Learner phase
In this phase, the Learners can enhance their knowledge via
interaction among themselves randomly. A learner learns new
things if the other learner has more knowledge than him or her.
The learning phenomenon of this phase at any iteration i for two
different learners Xi and Xj where ≠i j are given by
( ) ( )( )= + − < ( )X X r X X f X f X, If 23new i old i j i j i j, ,
( ) ( ) ( )= + − < ( )X X r X X f X f X, If 24new i old i j j i j i, ,
The pseudo-code of TLBO algorithm can be summarized as
follows:
Step 1: Determine the optimization problem in hand and in-
itialize the optimization parameters.
Step 2: Initialize the population (i.e. learners’) with random
generation and evaluate them.
Step 3: Choose the best learner of each subject as a teacher for
that subject and calculate mean result of learners in each
subject.
Step 4: Evaluate the difference between current mean result and
the best mean result according to Eq. (20) using the teaching
factor TF given in Eq. (21).
Step 5: Update the learners' knowledge by the help of teacher's
knowledge according to Eq. (22).
Step 6: Update the learners' knowledge by the knowledge of
some other learner according to Eqs. (23) and (24).
Step 7: Stop if a stopping criteria is achieved, else go to Step 3.
Fig. 5. Output of each individual IT2FLSs.

When a stopping criterion occurs, the result is the best answer
for the problem in hand (the best estimated parameters).
Generally speaking, in order to guide their search for optimum
solutions, heuristic algorithms, including TLBO, do not require any
information about the status of the system and control variables;
instead, they only need to check and measure the cost function
corresponding to the status of control variables. [36–40]. So, in this
paper, the cost function is considered as follows:
( )∑=
( )
CostFunction
N
e
1
25i
where ei is the trajectory error of ith sample for the object, N is the
number of sample and i is the iteration number. All parameters of
the GT2FPI are updated at every final time (tf). The General opti-
mization procedure for optimum General Type-II Fuzzy con-
trollers' coefficients determination with TLBO is presented in
Fig. 6.
As stated earlier, this paper proposes an online adaptive con-
troller using GT2FL and TLBO, for the objective of voltage regula-
tion in Power Systems. Fig. 7 displays the overall control frame-
work for online adjusting of membership functions of the fuzzy
rules, based on the TLBO technique.
Table 1 gives the applied optimal GT2FPI rules. The member-
ship functions corresponding to the input and output variables are
collocated as Negative Small (NS), Negative Medium (NM), Nega-
tive Large (NL), Positive Small (PS), Positive Medium (PM), and
Positive Large (PL). They have been collocated based on triangular
membership function which is the most famous and popular
membership function. The antecedent parts of each rule are
composed using AND function (with interpretation of minimum).
Here, Mamdani fuzzy inference system is also utilized. As will be
displayed in Section 6, the OGT2FPI controller has appropriate
efficiency in comparison to the classical techniques.
5. Advantages of the proposed method
Generally speaking, this paper proposes a novel kind of Fuzzy
Logic, General Type-II, in order to optimally tune parameters of a
Proportional-Integral (PI) controller. The main disadvantage of
type-2 fuzzy set is that type-2 fuzzy sets have three dimensions
and this new third dimension of type-2 fuzzy sets provides extra
degrees of freedom that facilitate control of uncertainties. To
overcome above mentioned problem, the paper introduces an
application of bio-inspired algorithms. Based on this, the authors
employ TLBO algorithm which is one of the recently proposed
population based algorithms. This algorithm does not require any
algorithm-specific control parameters. In the core of this method,
TLBO algorithm is utilized for evolving the parameters of the
controller and the parameters of input and output membership
functions. The proposed method is very simple. So, briefly it can be
written:
1. The proposed adaptive PI controller based on General type-2
fuzzy logic approach is easy to implement and can be applied to
a reasonably wide class of nonlinear systems.
2. The proposed method is based only on the available plant input/
output information and can be calculated on-line.
3. This proposed method is free of undesirable chattering phe-
nomena. Moreover, it can handle both structured and un-
structured uncertainties.
4. Another benefit of the suggested controller approach is its light
burden of computations which is an important figure in prac-
tical implementation and online control cases.
5. The proposed method is a free-model controller.
6. Simulation results
Fig. 8 illustrates the diagram of the case study system. In the
Fig. 6. General optimization procedure for optimum General Type-II Fuzzy con-
trollers' coefficients determination with TLBO.
Fig. 7. General scheme of proposed online TLBO-Fuzzy PI controller.
Table 1
The OGT2FPI rule set.
∆P ∆f
NL NM NS PS PM PL
S NL NM NS PS PS PM
M NL NL NM PS PM PM
L NL NL NL PM PM PM

case study system, a ±100MVAR STATCOM is installed having a 48-
pulse VSC that is connected to a 500-kV bus. This is the standard
sample STATCOM system in Matlab/Simulink library, and all ma-
chines used in the simulation are dynamical models [2,41].
Here, the attention of the analysis is focused on the control
performance of STATCOM in bus voltage regulation mode. It
should be noted that in the conventional model, in both voltage
and the current regulator, the regulation rate as well as the
amount of injection of the compensating reactive power are lar-
gely affected by parameters of the PI controller. The conventional
control scheme will be later compared to the proposed OGT2FPI
controller.
Assume that at the beginning, the steady-state voltage Vss
equals 1.0 p.u. As a disturbance, it is assumed that a voltage drop of
0.002 p.u. (from 1.0 to 0.989 p.u.) happens at the source (substa-
tion A) at time 0.2 s. It should be noted that, here, 0.989-p.u.
voltage at substation A actually denotes the lowest voltage that the
case study STATCOM device is able support because of its capacity
limit. After the fault is cleared, the voltage is assumed to gets high
back to about 1.0 p.u. In all simulations reported here, the STAT-
COM is assumed to start operating immediately right after the
disturbance, expecting that it will be able to bring the voltage back
to 1.0 p.u. The proposed control, OFPI, ∞H and the original PI
control are studied and compared.
6.1. Response of the Original Model
In the original model, Kp-V ¼ 12, Ki-V ¼ 3000, Kp-I ¼ 5, Ki-I¼ 40.
Here, all of the parameters are kept similar to the original model.
The initial voltage source, shown in Fig. 8, is 1 p.u., with the vol-
tage base being 500 kV. Observations are summarized in Table 2.
It can be inferred from Table 2 that that the proposed controller
is able to provide a quicker response compared to the original
conventional one. It should be noted that although necessary re-
active power amount is the same for both the proposed and ori-
ginal approaches, the proposed OGT2FPI approach responds faster,
as the voltage does. Fig. 9 shows the response of proposed control
method and H-Infinity with comparison with two other methods.
For the sake of simplicity, we may assume that the sensitivity
(ΔVar/ΔV) is a linear function. According to Fig. 10, when the
voltage error is 0.00001 p.u.Δ , Var is 0.0912MVar, which is in the
same range as the 0.12-MVar mismatch. Thus, it is reasonable to
conclude that the slight Var difference in Table 2 is due to round
off error in the dynamic simulation which always gives tiny ripples
beyond 5th digits even in the final steady state.
6.2. Change of PI control gains
In this scenario, the gains of the PI controller of the original
control are changed to Kp-V ¼ 1, Ki-V ¼ 1, Kp-I ¼ 1, Ki-I¼ 1 while all
other parameters of the system are kept the same as before.
The dynamic control gains depend on the post fault conditions;
It should be noted that they are independent of the initial condi-
tions of the system before the disturbance. Output results of ap-
plying the proposed methods based on proposed controller shown
in Fig. 10, respectively.
As illustrated in Fig. 11(a,c)., one can simply observe that when
the PI control gains of the original method are changed from their
original tuned values, the original control approach becomes un-
able to work properly and get the voltage back to 1 p.u., i.e. the
STATCOM demonstrates a poor response. More precisely, the re-
active power generated by the STATCOM is not enough to get the
voltage increased to a desirable value to meet the needs of the
system under disturbance. However, if the OGT2FPI control is
applied, the STATCOM becomes able to respond to the disturbance
Fig. 8. The general scheme of Case Study system.
Table 2
Performance comparison for the original system parameters.
Original Ctrl TLBO method
1
H-Infinity TLBO method
2
Lowest Voltage after
disturbance
0.9938p.u. 0.9938p.u. 0.9938p.u. 0.9938p.u.
Time(sec) when
V¼1.0
0.1695 s 0.1445 s 0.1308 s 0.1278 s
Var Amount at stea-
dy state
79.6MVar 79.46MVar 89.38Mvar 79.32MVar
Time to reach steady
state Var
0.1695 s 0.1445 s 0.1308 0.1278 s
Fig. 9. Output reactive power to the response of the Original Model.
Fig. 10. the voltage ripples of the system.

perfectly, i.e. the system voltage gets back to 1 p.u. quickly within
only 0.1 s. Fig. 11(b,d). also shows that the reactive power injection
cannot be continuously increased in the original control approach
to support system voltage, while the OGT2FPI control successfully
performs as desired.
6.3. Change of load
In this case, the original PI controller gains are kept the same as
originally designed, which means that Kp-V ¼ 12, Ki-V ¼ 3000,
Kp-I ¼ 5, Ki-I¼ 40.
However, the load at Bus B1 changes from 300 to 400 MW.
The results are shown in Fig. 12. It can be simply inferred that
the proposed method based on OGT2FPI can be successfully de-
signed and used to automatically react to a change in loads of the
system properly. Table 3 shows a few key observations of the
performance. From the data shown in Table 3 and Fig. 12, it is
obvious that the OGT2FPI control can achieve a quicker response
compared to the original approach.
6.4. Change of transmission network
In this case, the gains of PI controller again remain unchanged,
i.e. they are similar to the original design. However, line 1 is
switched off at 0.2 s to represent a change in the grid which may
(c)
(a) (b)
OGT2FPI
(d)
Fig. 11. The effect of changes PI control gains on output reactive power and Voltages.
Fig. 12. The effect of load changes on output reactive power.
Table 3
Performance comparison with a change of load.
Original Ctrl TLBO method
1
H-Inﬁnity TLBO method
2
disturbance
0.9949p.u. 0.9949p.u. 0.9949p.u. 0.9949p.u.
Time(s) when V¼1.0 0.1695 s 0.1445 s 0.1308 s 0.1278 s
Var Amount at stea-
dy state
93.08MVar 92.95MVar 0.9277Var 92.67MVar
Time to reach steady
state Var
0.1695 s 0.1445 s 0.1308 s 0.1278 s
Fig. 13. The effect of a change of transmission network.

correspond to scheduled transmission maintenance.
The results are shown in Fig. 13. Key observations are sum-
marized in Table 4. Observations support the idea that the pro-
posed method based on OGT2FPI can be designed to automatically
react to changes in the transmission network.
It should be noted that, in this case, the STATCOM absorbs VAR
from the power system. Here, the disturbance should cause a
voltage rise at (substation A) from 1.0 to 1.01 p.u.; however, on the
other hand, due to loss of a line in the grid, the voltage tends to
drop. If the STATCOM is not activated, the combinational impact
causes the steady state voltage to rise to higher than 1.0 p.u. at the
controlled bus As a result of overvoltage, the STATCOM should
absorb VAR in the final steady state so that 1.0 p.u. voltage is ob-
tained at the controlled bus. As shown in Fig. in case of using the
original control method, initial transients immediately after 0.2 s
has led to an undesired over absorption of reactive power by the
STATCOM; while, on the other hand, when the OGT2FPI control is
used, a much smoother and quicker response is obtained.
6.5. Summary of the simulation study
It can be simply inferred from the abovementioned case studies
that the proposed method based on OGT2FPI control can achieve
faster and more consistent response compared to the original
control approach. Interestingly, the response time and the output
curve of the proposed OGT2FPI control are almost identical under
various disturbances such as a change in (initial) control gains, a
change in load, and also a change in network topology. In contrast,
when the original control method is applied, the response curve of
the system varies greatly under different change of system oper-
ating condition and more importantly, the STATCOM may some-
time not be able to modify the voltage to the desired value. Of
course, it was already expected that the proposed controller would
outperform the original control method as it dynamically and
autonomously adjusts controller's gains during the voltage cor-
rection process; therefore, the desired performance can be
achieved adaptively.
Remark 1. The target audience of this paper is not only intended
for students and researchers but will also be valuable for execu-
tives, managers, marketing experts and project leaders who would
like to apply fuzzy logic control to industrial applications such as
the control of power systems and electronics, Robotics and so on.
The manuscript presents a set of new fuzzy methods, case studies
and optimization algorithm which together make it possible to
improve the efficiency and effectiveness of PI control.
7. Conclusion
This paper presents a novel online intelligent strategy for
controlling Voltage Regulation in Power Systems. The proposed
framework applies a combination of a simple heuristic algorithm,
named Teaching Learning Based Optimization (TLBO), and a new
General Type-II Fuzzy Logic to optimally and adaptively tune gains
of a proportional-integral controller which is used widely in
STATCOM devices. Conventionally, implementing general type-2
fuzzy systems imposes high computational burden; however, by
using a recently introduced plane representation, GT2FS can be
regarded as a combination of several interval type-2 fuzzy logic
systems (IT2FLS) each one of which has its own corresponding
level of α. Linguistic rules are directly incorporated into the con-
troller. In addition, an H1 compensator is corrupted to attenuate
external disturbance and fuzzy approximation error. The proposed
control methodology is evaluated on a test case power system
owning a ±100-MVAR STATCOM with a 48-pulse VSC and con-
nected to a 500-kV bus, which represents a large power system.
Simulation results indicate that the simple TLBO algorithm is able
to successfully determine optimal parameters of the controller as
well as the parameters of input and output membership functions.
The proposed technique is straightforward and free of computa-
tional complexity. To evaluate the performance of the proposed
framework, simulation results of the proposed method are com-
pared with those of the conventional PI controller and OFPI con-
troller, which are the latest researches in the problem in question.
Results indicate the superiority of the proposed framework.
Acknowledgements
Appreciation is offered to the Iran National Science Foundation
for a grant to collect data for this research. The authors wish to
thank the Iran National Science Foundation for supporting helps to
do this research and providing this opportunity for the survey
team. Moreover, the authors would like to thank the reviewers for
their very useful comments, which helped us to improve the
quality of the paper. The reviewers’ efforts are gratefully
appreciated.
References
[1] Aghaei J, Gitizadeh M, Kaji M. Placement and operation strategy of FACTS
devices using optimal continuous power flow. Sci Iran 2012;19(6):1683–90.
[2] Xu Yao, Fangxing Li. Adaptive PI control of STATCOM for voltage regulation.
Power Deliv, IEEE Trans 2014;29(3):1002–11.
[3] Wang Keyou, Crow Mariesa L. Power system voltage regulation via STATCOM
internal nonlinear control. Power Syst, IEEE Trans 2011;26(3):1252–62.
[4] Schauder, Colin, and Hars‘nadMehta. Vector analysis and control of advanced
static VAR compensators. In: IEEE Proceedings C (Generation, Transmission
and Distribution), vol. 140, 4, pp. 299–306. IET Digital Library, 1993.
[5] Lehn PW, Iravani MR. Experimental evaluation of STATCOM closed loop dy-
namics. Power Deliv, IEEE Trans 1998;13(4):1378–84.
[6] Rao Pranesh, Crow ML, Yang Zhiping. STATCOM control for power system
voltage control applications. Power Deliv, IEEE Trans 2000;15(4):1311–7.
[7] Dong Liangying, Crow ML, Yang Z, Shen C, Zhang L, Atcitty S. A reconfigurable
FACTS system for university laboratories. Power Syst, IEEE Trans 2004;19
(1):120–8.
[8] El-Moursi MS, Sharaf AM. Novel controllers for the 48-pulse VSC STATCOM
and SSSC for voltage regulation and reactive power compensation. Power Syst,
IEEE Trans 2005;20(4):1985–97.
[9] Saeedifard Maryam, Nikkhajoei Hassan, Iravani Reza. A space vector modu-
lated STATCOM based on a three-level neutral point clamped converter. Power
Deliv, IEEE Trans 2007;22(2):1029–39.
[10] Hatano Nobuhiko, Ise Toshifumi. Control scheme of cascaded H-bridge STAT-
COM using zero-sequence voltage and negative-sequence current. Power
Deliv, IEEE Trans 2010;25(2):543–50.
[11] Wang H_F, Li H, Chen H. Application of cell immune response modelling to
power system voltage control by STATCOM. IEE Proc-Gener, Transm Distrib
2002;149(1):102–7.
[12] Chen Woei-Luen, Hsu Yuan-Yih. Controller design for an induction generator
driven by a variable-speed wind turbine. Energy Convers, IEEE Trans 2006;21
(3):625–35.
[13] Spitsa Vitaly, Alexandrovitz Abraham, Zeheb Ezra. Design of a robust state
feedback controller for a STATCOM using a zero set concept. Power Deliv, IEEE
Trans 2010;25(1):456–67.
[14] Luo An, Tang Ci, Shuai Zhikang, Tang Jie, Xu Xian Yong, Chen Dong. Fuzzy-PI-
based direct-output-voltage control strategy for the STATCOM used in utility
distribution systems. Ind Electron, IEEE Trans 2009;56(7):2401–11.
Table 4
Performance comparison with changed transmission.
Original Ctrl TLBO method 1 TLBO method 2
disturbance
0.9949p.u. 0.9949p.u. 0.9949p.u.
Time(s) when V¼1.0 0.1695 s 0.1445 s 0.1278 s
Var Amount at steady state 93.08MVar 92.95MVar 92.67MVar
Time to reach steady state
Var
0.1695 s 0.1445 s 0.1278 s

[15] Rahim AHMA. Fuzzy STATCOM control strategies for power system stabiliza-
tion. ICGST Int J Autom Control Syst Eng, ACSE 2006;6(2):41–8.
[16] Mak LO, Ni YX, Shen CM. STATCOM with fuzzy controllers for interconnected
power systems. Electr Power Syst Res 2000;55(2):87–95.
[17] Liu Chien-Hung, Hsu Yuan-Yih. Design of a self-tuning PI controller for a
STATCOM using particle swarm optimization. Ind Electron, IEEE Trans 2010;57
(2):702–15.
[18] Khooban MH, Niknam T, Blaabjerg F, Davari P, Dragicevic T. A robust adaptive
load frequency control for micro-grids. ISA Trans 2016. http://dx.doi.org/
10.1016/j.isatra.2016.07.002.
[19] Khooban MH, Niknam T, Blaabjerg F, Dehghani M. A free chattering hybrid
sliding mode control for a class of nonlinear systems: electric vehicles as a
case study. IET Sci, Meas Technol 2016. http://dx.doi.org/10.1049/
iet-smt.2016.0091.
[20] Naghash-Almasi O, Khooban MH. PI adaptive LS-SVR control scheme with
disturbance rejection for a class of uncertain nonlinear systems. Eng Appl Artif
Intell 2016;52:135–44.
[21] Khooban MH, Niknam T, Sha-Sadeghi M. Speed control of electrical vehicles: a
time-varying proportional–integral controller-based type-2 fuzzy logic. IET Sci,
Meas Technol 2016;10(3):185–92.
[22] Khooban MH, Naghash-Almasi O, Niknam T, Sha-Sadeghi M. Intelligent robust
PI adaptive control strategy for speed control of EV (s). IET Sci, Meas Technol
2016;10(5):433–41.
[23] Khooban Mohammad, Hassan Davood Nazari Maryam, Abadi Alireza, Alfi,
Siahi Mehdi. Optimal Type-2 fuzzy controller for HVAC systems. AUTO-
MATIKA: Časopis za Autom, Mjer, Elektron, računarstvo i Komun 2014;55
(1):69–78.
[24] Chiclana F, Zhou S-M. Type-reduction of General Type-2 Fuzzy Sets: the Type-
1 OWA approach. Int J Intell Syst 2013;28(5):505–22.
[25] Ghaemi Mostafa, Hosseini-Sani Seyyed Kamal, Khooban Mohammad Hassan.
Direct adaptive general type-2 fuzzy control for a class of uncertain non-linear
systems. IET Sci, Meas Technol 2014;8(6):518–27.
[26] Zadeh LA. The concept of a linguistic variable and its application to approx-
imate reasoning—I. Inf Sci 1975;8(3):199–249.
[27] Khooban, Mohammad Hassan, and TaherNiknam. A new and robust control
strategy for a class of nonlinear power systems: Adaptive general type-II fuzzy.
Proceedings of the Institution of Mechanical Engineers, Part I: Journal of
Systems and Control Engineering (2015): 0959651815571621.
[28] Liu F. An efficient centroid type-reduction strategy for general type-2 fuzzy
logic system, IEEE Comput Intell Soc, Walter J Karplus Summer Res Grant Rep,
pp; 2006.
[29] Shahsadeghi Mokhtar, Khooban Mohammad Hassan, Niknam Taher. A robust
and simple optimal type II fuzzy sliding mode control strategy for a class of
nonlinear chaotic systems. J Intell Fuzzy Syst: Appl Eng Technol 2014;27
(4):1849–59.
[30] Liu F. An efficient centroid type-reduction strategy for general type-2 fuzzy
logic system. Inf Sci 2008;178(9):2224–36.
[31] Karnik, N.N., Mendel, J.M.Type-2 fuzzy logic systems: type-reduction. IEEE
International Conference on Systems, Man, and Cybernetics, 1998, 11–14 Oc-
tober 1998, vol. 2042, pp. 2046–2051.
[32] Karnik NN, Mendel JM. Centroid of a type-2 fuzzy set. Inf Sci 2001;132(1–
4):195–220.
[33] Karnik NN, Mendel JM, Qilian L. Type-2 fuzzy logic systems. IEEE Trans Fuzzy
Syst 1999;7(6):643–58.
[34] Rao Ravipudi V, Savsani Vimal J, Vakharia DP. Teaching–learning-based opti-
mization: a novel method for constrained mechanical design optimization
problems. Comput-Aided Des 2011;43(3):303–15.
[35] Rao RV, Savsani VJ, Vakharia DP. Teaching–learning-based optimization: an
optimization method for continuous non-linear large scale problems. Inf Sci
2012;183(1):1–15.
[36] Khalghani MR, Khooban MH, Mahboubi-Moghaddam E, Vafamand N, Good-
arzi M. A self-tuning load frequency control strategy for microgrids: human
brain emotional learning. Int J Electr Power Energy Syst 2016 29;75:311–9.
[37] Khalghani Mohammad Reza, Khooban Mohammad Hassan. A novel self-tun-
ing control method based on regulated bi-objective emotional learning con-
troller's structure with TLBO algorithm to control DVR compensator. Appl Soft
Comput 2014;24:912–22.
[38] Mahboubi-Moghaddam E, Narimani MR, Khooban MH, Azizivahed A. Multi-
objective distribution feeder reconfiguration to improve transient stability,
and minimize power loss and operation cost using an enhanced evolutionary
algorithm at the presence of distributed generations. Int J Electr Power Energy
Syst 2016;76:35–43.
[39] Khooban MH, Vafamand N, Niknam T. T–S fuzzy model predictive speed
control of electrical vehicles. ISA Trans 2016. http://dx.doi.org/10.1016/j.
isatra.2016.04.019.
[40] Modirkhazeni A, Almasi ON, Khooban MH. Improved frequency dynamic in
isolated hybrid power system using an intelligent method. Int J Electr Power
Energy Syst 2016;78:225–38.
[41] Matlab & Simulink, GTO-based STATCOM Dec. [Online]. Available: 〈http://
www.mathworks.com/help/physmod/sps/pow-ersys/ug/gto-based-statcom.
html〉, Feb. 2012; 2013.

Adaptive PI Controller for Voltage Regulation in Power Systems

More Related Content

What's hot (20)

Similar to Adaptive PI Controller for Voltage Regulation in Power Systems (20)

More from ISA Interchange (20)

Recently uploaded (20)

Adaptive PI Controller for Voltage Regulation in Power Systems