You are on page 1of 14

Journal of Electrical Engineering & Technology

https://doi.org/10.1007/s42835-020-00386-w

ORIGINAL ARTICLE

Swarm Intelligence Approach for Angle Stability Improvement of PSS


and SVC‑Based SMIB
Nor Azwan Mohamed Kamari1 · Ismail Musirin2 · Ahmad Asrul Ibrahim1

Received: 22 October 2019 / Revised: 21 January 2020 / Accepted: 18 February 2020


© The Korean Institute of Electrical Engineers 2020

Abstract
This paper discusses the effect of multi-objective function in angle stability improvement for a single machine connected to
an infinite bus system (SMIB). Minimum damping ratio, ξmin and maximum damping factor, σmax which are commonly used
to indicate oscillatory stability condition in power system are merge in certain ratio to produce a multi-objective function,
FMO. This new index brings the advantages of the two indices without compromising the weakness of the index involved.
In this study, FMO is applied to tune parameters of static var compensator with proportional-integral-derivative controller
(SVC-PID) to improve damping efficiency in SMIB. The result is compared with a system connected to power system stabi-
lizer attached with lead lag controller (PSS-LL). The parameters of SVC-PID and PSS-LL are optimized by particle swarm
optimization method. Validation based on speed response, phase plane and determination of eigenvalues confirms that the
proposed FMO is more effective for solving angle stability problems compared to single objective function.

Keywords Angle stability · Damping ratio · Power system stabilizer · Static var compensator · Particle swarm optimization

1 Introduction multi-objective function, FMO is required to address the


holistic improvement in the system. Among the well-known
Due to the rising fuel and manpower costs, the manage- indicators in determining angle stability are the damping
ment of power generation has always been a priority over ratio and damping factor. Using minimum damping ratio
the world. With high restrictions on spending, the power ξmin as an index, the optimization of this indicator can move
system needs to be well managed so that system stability can the coordinates of each eigenvalue in the complex s-plane
be maintained and not lethal due to any disruption [1–20]. approaching the real (σ) axis lines [3–6]. While the maxi-
Indicator of the power system is often used as a determi- mum damping factor σmax can give better results in moving
nant in justifying the stability of the system. However, only the eigenvalue coordinates far to the left of the imaginary
single indicators are often used during the determination (jω) axis [7]. The combination of these two indices may able
process for power system stability tuning. This causes the to bring the advantage without compromising the justifica-
system does not succeed in improving the overall quality tion of one of the indexes.
[5–8]. Therefore, a combination of some indicators to create Improvements in angle stability is often a major focus
in the management of the power system stability. Imple-
mentation of the damping devices to the system are nec-
* Nor Azwan Mohamed Kamari
azwank@ukm.edu.my essary so that the stability of the system can be restored
or improved. Power system stabilizer (PSS) is usually the
Ismail Musirin
ismailbm@salam.uitm.edu.my preferred choice to be used to increase the capacity of the
power system damping [9–12]. Apart from PSS, flexible cur-
Ahmad Asrul Ibrahim
ahmadasrul@ukm.edu.my rent transmission (FACTS) devices are also widely used,
namely static var compensator (SVC) [13–16], compensa-
1
Department of Electrical, Electronic and Systems tion series control thyristor (TCSC) [17–19], static synchro-
Engineering, Faculty of Engineering and Built Environment, nous compensator (STATCOM) [20–22], static synchronous
Universiti Kebangsaan Malaysia, Bangi, Malaysia
series compensator (SSSC) [23–25] and unified power
2
Faculty of Electrical Engineering, Universiti Teknologi flow controller (UPFC) [26–28]. SVC is a combination of
Mara, Shah Alam, Malaysia

13
Vol.:(0123456789)
Journal of Electrical Engineering & Technology

thyristor-controlled regulator (TCR) and fixed capacitor the searcher ants to solve the calculation problem in track-
(FC). On the other hand, TCSC can provide sufficient load ing the best route. Meanwhile, PSO concept can be said to
compensation by controlling the reactance of a transmission mimic the movement of a herd, such as the behavior of fish
line effectively. Meanwhile, STATCOM can be described schooling and insects swarming. This technique was origi-
as a second generation of shunt connected FACTS device nally founded based on the population of random particles
based upon voltage source converter (VSC). STATCOM in which every particle is a potential solution. The PSO is
supply required reactive power to maintain bus voltage and able to make adjustments so that there is a balance between
improves damping capability of the system. In spite of both global and local exploration during the search process. This
SVC and STATCOM have the significant capability, the feature makes the PSO more confident in overcoming the
installation cost of STATCOM is higher compared to SVC. problems caused by initial convergence. it also improves
SSSC is a combination of voltage source converter (VSC) the ability to search.
in solid state and transformer. Both are connected in series This paper discussed the effectiveness of multi objec-
with the transmission line and contribute sinusoidal voltage tive function, FMO in angle stability improvement for a sin-
with variable magnitude to the line. In this paper, two damp- gle machine connected to an infinite bus system (SMIB).
ing controllers, power system stabilizer (PSS) and static var Based on the efficiency of FMO, the performance of static
compensator (SVC) are compared. PSS and FACTS devices var compensator with PID controller (SVC-PID) to improve
are often installed together with controllers such as lead- damping efficiency in SMIB is compared to power system
lag (LL) [29–31], proportional-integral-derivative (PID) stabilizer that attached with lead lag controller (PSS-LL).
[32–34] and fuzzy [35–37] to strengthen the damping effi- The achievement of FMO, minimum damping ratio ξmin and
ciency of the system. In this study, SVC will be attached to maximum damping factor, σmax in tuning damping control-
PID controller, while PSS is combined with LL controller. lers are also been analyzed. Fixed-gains of both SVC-PID
In recent years, the use of artificial intelligence (AI) tech- and PSS-LL are justified using Particle Swarm Optimization
nology has become the preferred choice in solving power (PSO) method. For optimization technique comparisons,
system problems. Unstable power systems require equipment SVC-PID optimized with PSO method is compared to SVC-
such as power system stabilizers, flexible AC transmission PID optimized with EP and AIS techniques. The goal of this
systems (FACTS), PID controllers and LL controllers to study is to produce a stable SMIB system in minimum time.
increase stability. This device, especially the PID and LL
controllers has various parameters that need to be tuned.
Here, the use of AI is introduced to adjust the parameters 2 SMIB System
involved. The AI approach is applied to solve optimum
values in a system or condition, particularly in the field of The power system model tested in this study is based on a
economic dispatch, capacitor placement and sizing, and single machine connected to an infinite bus system (SMIB),
assessment and improvement of voltage and angle stability. as shown in Fig. 1. This concept can be symbolized as a
Among the AI approaches commonly used in power sys- synchronous machine connected by a transmission line with
tems are evolutionary computation (EC) and artificial neu- a large bus that has an infinite load [4, 10]. Re and Xe, are the
ral networks (ANN) [38–40]. EC field include evolutionary real and imaginary parts of the transmission line impedance,
programming (EP) [41–43], particle swarm optimization respectively.
(PSO) [44–46], artificial immune systems (AIS) [47–49], For a deeper analysis, a Philips-Heffron model has been
whale optimization algorithm (WOA) [15, 50, 51] and ant developed based on the SMIB system model in Fig. 1. A
colony optimization (ACO) [52–54]. EP algorithm is mod- block diagram comprising the Phillips–Heffron model for
eled on the biological evolution process in solving a complex SMIB system shown in Fig. 2.
problem. The main features of the EP are process of muta- H is the constant of inertia. KR is the exciter system’s
tion of the next generation and selection of more powerful circuit constant. Kd is the coefficient of damping torque. TR
genes. AIS algorithm can be said to use a concept similar
to and EP. While both concepts are biologically based on
living things, EP focuses on the evolution of living things,
while AIS adopts the concept of living immune system. The
difference between AIS and EP algorithms is that AIS has
an additional process of cloning called the clonal selection
algorithm (CSA). On the other hand, the ACO approach is
inspired by the true behavior of ants while searching for food
and interacting with fellow ants. In ACO, artificial ants (the Fig. 1  A single machine connected to an infinite bus system (SMIB)
search agent) will communicate using pheromone. It guides model

13
Journal of Electrical Engineering & Technology

Fig. 3  Philips–Heffron block diagram model for PSS-LL

obtained from the gain block. Meanwhile, two phase-com-


pensation blocks contribute two level of relevant phase-lead
Fig. 2  Phillips–Heffron block diagram model for SMIB characteristic [9, 12, 23].
T1 and T2 are the time constants for the first phase com-
is the exciter system’s time constant. ω0 = 2πf0. K1–K6 and pensation. T3 and T4 are the time constants for the second
T3 are constants associated with electric torque, rotor speed, phase compensation. Tw is the washout time constant. Kst is
rotor angle, ratio of the impedance, load and level of excita- the stabilizer gain for PSS. In this study, both phase compen-
tion. Referring to Fig. 2, the equations representing SMIB sation blocks were determined to have the same phase, then
Phillips–Heffron system model can be derived as follow: T1 = T3 and T2 = T4. So that, the simulation test for SMIB
system with PSS-LL, three parameters: T1, T2 and Tw which
Δ𝜔 ΔTm − K1 Δ𝛿 − Kd Δ𝜔r − K2 ΔEq need to be optimized. All parameter values for PSS-LL used
= (1)
Δt 2H in this study are based on the parameters recorded in [9].

Δ𝛿
= 𝜔0 Δ𝜔r (2)
2.2 SVC‑PID
Δt
In this paper, static var compensator with PID controller
ΔEq K3 K4 Δ𝛿 + ΔEq − K3 Δ𝜈f (SVC-PID) is proposed as a new design in improving stabil-
=− (3) ity in SMIB. The SVC-PID can be separated into two blocks:
Δt TK
block of PID controller and block of SVC. The Philips–Hef-
fron block diagram model for SVC-PID is shown in Fig. 4
Δ𝜈f KR K5 Δ𝛿 + KR K6 ΔEq + Δ𝜈f + KR Δ𝜎
=− (4) [15, 26].
Δt TR KV is the SVC circuit constant. T V is the SVC time con-
stant. KP, KI and KD are the proportional, integral and deriv-
where Tm is a mechanical torque. Equations (1)–(4) can be
ative gain of the PID controller, respectively. For SMIB sys-
encompassed into a matrix form as followed:
tem with SVC-PID, these three parameters are considered to
Ẋ = A ⋅ X + B ⋅ U (5) be optimized. All parameter values for SVC-PID used in this
study are based on the parameters recorded in [34].
[ ]T
X= Δ𝜔r Δ𝛿 ΔEq Δvf Δ𝜎 Δ𝛽 (6)

Comprehensive calculation of the system in Fig. 2 can be 3 Optimization Techniques


found in [21]. All parameter values for generators, transmis-
sion lines and exciters used in this study are based on the In this study, three optimization techniques namely, particle
parameters recorded in [34]. swarm optimization (PSO), evolutionary programming (EP)
and artificial immune system (AIS) methods are applied to
search the optimal value of parameters of SVC-PID and
2.1 PSS‑LL PSS-LL that attached to SMIB.

In this study, the SMIB model shown in Fig. 2 will be con-


nected to the PSS with the lead lag controller (PSS-LL). The
PSS-LL can be portrayed as three types of blocks that con-
nected in series; ie a washout block, a gain block, and two
phase-compensation blocks. Figure 3 shows the Philips–Hef-
fron block diagram model for PSS-LL. The gain block gain
regulates the damping signal received from the PSS input,
while the washout block serves as a high pass filter for input Fig. 4  Philips–Heffron block diagram model for SVC-PID

13
Journal of Electrical Engineering & Technology

3.1 PSO certain velocity, the particles fly through space by follow-


ing optimum particles. Then, it seeks optimum solution by
Particle swarm optimization (PSO) was introduced in 1995 updating the position and velocity of each particle. There
by Kennedy and Eberhart. PSO is one of the evolution- are several stages in PSO algorithm such as initialization,
based optimization techniques. The PSO concept can be velocity updates, position updates, fitness calculations and
said to mimic the movement of a herd, such as the behavior best-ranked updates. Figure 5 represented the flowchart for
of birds flocking and insects swarming. This technique was the PSO algorithm. Comprehensive explanations of PSO
originally founded based on the population of random par- algorithm process can be found in [26, 34].
ticles in which every particle is a potential solution. With
3.2 EP

In the evolutionary programming (EP) algorithm, the pro-


cesses involved are initialization, statistical evaluation,

Fig. 5  Flowchart for the particle swarm optimization (PSO) algorithm


Fig. 8  The sector of eigenvalues on a complex s-plane (before and
after the optimization process of ξmin)

Fig. 9  The sector of eigenvalues on a complex s-plane (before and


after the optimization process of σmax)

Fig. 6  Flowchart for the evolutionary programming (EP) algorithm

Fig. 10  The sector of eigenvalues on a complex s-plane (before and


after the optimization process of FMO)

Table 1  Loading conditions for Case 1, Case 2 and Case 3


Loading condition Case 1 Case 2 Case 3

Active power, P (p.u.) 0.5 0.35 − 0.15


Reactive power, Q (p.u.) 0.15 − 0.15 0.3
Fig. 7  Flowchart for the artificial immune system (AIS) algorithm

13
Journal of Electrical Engineering & Technology

Fig. 11  Speed response and


eigenvalues for Case 1

13
Journal of Electrical Engineering & Technology

Table 2  Correlation of all seven approaches for Case 1 on the complex s-plane will move to the left of the imagi-
Type KP KI KD nary (jω) axis regularly. The optimization result of ξmin also
decreases the imaginary value of each eigenvalues. This
SVC-PID causes the sector of eigenvalues to move closer to the real
PID-MO 0.5236 − 0.0995 0.2532 (σ) axis. The effect of the optimization of ξmin will form a tri-
PI-MO 0.5468 0.6310 – angular-shaped sector of eigenvalues in the complex s-plane.
PI-Zt 0.4612 0.2405 – Figure 8 shows the sector of eigenvalues that mapped on a
PI-Sg 0.3215 0.1905 – complex s-plane. White and black square are the eigenvalues
Type T1(= T3) T2(= T4) TW location before and after the optimization process of ξmin,
respectively.
PSS-LL
PSS-MO 0.0900 9.0139 3.9365
Whereas, the maximum damping factor σmax calculates
PSS-Zt 0.5603 1.3656 1.8641
the maximum value (or the minimum of the negative value)
PSS-Sg 1.7466 1.6337 1.0695
damping factor σ for each eigenvalue, as follows [7]:
( )
𝜎max = max 𝜎i (8)

From σmax optimization, the eigenvalue coordinates are


fitness computation, mutation, combination and selection.
moved far to the left from the jω axis. The sector of eigen-
Figure 6 illustrated the flowchart of EP algorithm process.
values that mapped on a complex s-plane, before and after
the optimization process of σmax are shown in Fig. 9.
3.3 AIS
By implementing the FMO optimization, this will optimize
both ξmin and σmax at the same time, which are to move the
Artificial immune system (AIS) algorithm that involved
sector of eigenvalues to the left of the jω axis and at the
are almost the same with EP, except one extra component,
same time move closer to the σ axis. The sector of eigenval-
namely cloning. Figure 7 illustrated the flowchart of AIS
ues that mapped before and after the optimization process of
algorithm process.
FMO on a complex s-plane are shown in Fig. 10. With FMO
optimization, the coordinates of eigenvalues can be repre-
sented as being within the boundary of the trapezium-shaped
4 Multi‑objective Function for Angle
sector in the complex s-plane.
Stability Improvement
In this study, the minimum damping ratio ξmin and maxi-
mum damping factor σmax are collaborated into a multi-
This paper proposes a multi-objective function (FMO) to
objective function FMO equation as follows:
reduce the deviation of the rotor speed and at the same time
enhances the damping ability in the power system. FMO FMO = 𝜌1 ⋅ 𝜁min + 𝜌2 ⋅ 𝜎max , 0 ≤ 𝜌1 , 𝜌2 ≤ 1.0 (9)
was obtained based on a combination of two conventional
indexes in angular stability ie damping factor and damping ρ1 and ρ2 are the constants attached to ξmin and σmax, respec-
ratio. Only the minimum value of the damping ratio, ξmin tively. It is used to determine the percentage of both indica-
and the maximum value of the damping factor, σmax used in tors. ωi and σi are respectively the imaginary and real parts
FMO according to a certain ratio. In this study, SVC-PID and of the ith eigenvalue. FMO can be formulated as:
PSS-LL have been selected as damping controllers added to Maximum(FMO ) (10)
the system. Only one damping controller will be used at a
time. The selected damping controller will be tuned, and ξmin FMO has been applied to tune parameters of SVC-PID as
and σmax values will be calculated. reported in [8], with different loading conditions compared
The calculation of ξmin is based on the real value σ and to this study. The result gives the best ratio of minimum
the imaginary value ω of each eigenvalue in the system, as damping ratio (ξmin) and maximum damping factor (σmax) to
follows [4]: tune parameters in SVC-PID.
Here, the limitation of washout time constant, T W, the first
⎛ ⎞ phase-compensation time constant, T1 and the second phase-
⎜ 𝜎k ⎟
𝜁min = min⎜ � ⎟ (7) compensation time constant, T3 values need to be considered
⎜ 𝜎 2 + 𝜔2 ⎟ for systems injected by the PSS-LL:
⎝ k k⎠
max min
TW ≤ TW ≤ TW , T1max ≤ T1 ≤ T1min , T3max ≤ T3 ≤ T3min
Here, k represents the number of eigenvalues. As a result
(11)
of ξmin optimization, the coordinates of each eigenvalues

13
Journal of Electrical Engineering & Technology

Fig. 12  Speed response and


eigenvalues for Case 2

13
Journal of Electrical Engineering & Technology

Table 3  Correlation of all seven approaches for Case 2 PSS-Sg, in which all the three systems have been damped
Type KP KI KD within 3 s. From the oscillation perspective, PID-MO also
gives the smallest oscillation rate compared to the other six
SVC-PID systems. All three systems tuned with SVC-PI controller are
PID-MO 0.5018 − 0.1205 0.1803 among the worst, with PI-Sg gives the biggest oscillation
PI-MO 0.5467 0.6528 – rate and takes the longest time to damp. From the result of
PI-Zt 0.5421 0.3724 – Case 1, it is clearly shown that PID-MO is the most capable
PI-Sg 0.2710 0.0299 – in enhancing the best damping performance of SMIB system
Type T1(=T3) T2(=T4) TW compared to other six approaches. It is also shown that among
the systems with same installed devices, the FMO approach is
PSS-LL
more capable to produce shorter damping time and smaller
PSS-MO 0.0716 3.9281 0.7441
oscillation rate compared to the systems with ξmin and σmax
PSS-Zt 1.4970 1.2500 0.7898
approaches.
PSS-Sg 0.4910 2.5497 4.8348
Figure 11b shows the phase plane response for Case 1.
From the result, PID-MO stopped in 2 cycles, the smallest
number of cycles compared to others. Meanwhile, all the three
For systems with SVC-PID, here are the limitations of PSS-LL systems stopped in the range of 5–7 cycles. On the
values for proportional gain, KP, integral gain, KI and deriva- other hand, PI-MO, PI-Zt and PI-Sg methods are among the
tive gain, KD: worst, which take more than eight cycles before the phase-
plan revolution finished. Based on the phase plan perspective,
KPmax ≤ KP ≤ KPmin , KImax ≤ KI ≤ KImin , KDmax ≤ KD ≤ KDmin the PID-MO approach is in the first position for improving
(12) the damping performance of SMIB system. The result also
indicates that under the same installed devices such as PSS-
LL or SVC-PI controller, the MO method can give far more
5 Results and Discussion accurate tuning of damping controller and give better enhance-
ment of stability performance compared to ξmin and σmax for
In this study, the performances of SMIB system that attached SMIB system.
with SVC-PID using multi-objective function FMO approach Figure 11c shown the region of eigenvalues in complex
(PID-MO) are compared to SMIB with SVC-PI using FMO s-plane for Case 1. All eigenvalues are scattered at the nega-
approach (PI-MO), SMIB with SVC-PI using ξmin approach tive side of the real axis which justified that all results are in
(PI-Zt), SMIB with SVC-PI using σmax approach (PI-Sg), stable condition. From all the seven techniques, the eigenval-
SMIB with PSS-LL using FMO approach (PSS-MO), SMIB ues of PID-MO are positioned at the most left of negative
with PSS-LL using ξmin approach (PSS-Zt) and SMIB with side, as well as at the nearest to the real axis of the complex
PSS-LL with σmax approach (PSS-Sg). Also, the performance s-plane. This result shows that PID-MO has more capability
of both PI-MO and PSS-MO are compared to SVC-PI sys- to enhance the stability level of SMIB system as compared to
tem with the ξmin approach (PI-Zt), SVC-PI system with the other methods. On the other hand, one of the eigenvalues of
σmax approach (PI-Sg), PSS-LL system with the ξmin approach PI-Zt is at the nearest location to the origin. This verified that
(PSS-Zt) and PSS-LL system with the σmax approach (PSS- PI-Zt approach is at the lowest level of stable condition among
Sg). To present the above comparison events, three cases with all seven techniques. Among other approaches, eigenvalues of
various loading conditions are simulated. Table 1 tabulated PSS-Sg are located at the furthest from imaginary axis of com-
the loading conditions for all three cases: Case 1, Case 2 and plex s-plane. This technique is unable to improve ξmin value
Case 3. that leads to shift the eigenvalues nearer to the real axis of
In this study, all cases are using PSO as the computational complex s-plane. The value of SVC-PID, SVC-PI and PSS-
intelligence technique. For system with the FMO approach, LL tuned by all seven techniques: PID-MO, PI-MO, PI-Zt,
both constant values attached to ξmin, ρ1 and σmax, ρ2 both are PI-Sg, PSS-MO, PSS-Zt and PSS-Sg for Case 1 are tabulated
set to 0.5, respectively. These values are randomly chosen in in Table 2.
the range [0,1]. The responses of speed deviation, phase plane and the
The responses of speed deviation, phase plane and the region of eigenvalues in complex s-plane comparing all
region of eigenvalues location in complex s-plane comparing the seven methods for Case 2 are shown in Fig. 12a–c,
PID-MO, PI-MO, PI-Zt, PI-Sg, PSS-MO, PSS-Zt and PSS-Sg respectively.
for Case 1 are shown in Fig. 11a–c, respectively. From the speed responses, the performances of all tech-
From Fig. 11a, the PID-MO system takes the shortest niques are improved drastically than the previous one. The
damping time, about 2 s; followed by PSS-MO, PSS-Zt and speed responses for all techniques were damped within

13
Journal of Electrical Engineering & Technology

Fig. 13  Speed response and


eigenvalues for Case 3

13
Journal of Electrical Engineering & Technology

Table 4  Correlation of all seven approaches for Case 3


Type KP KI KD

SVC-PID
PID-MO 0.6029 − 0.0903 0.3163
PI-MO 0.5503 0.5838 –
PI-Zt 0.6400 0.6470 –
PI-Sg 0.1507 0.3174 –
Type T1(=T3) T2(=T4) TW

PSS-LL
PSS-MO 0.0987 0.2592 1.2129
PSS-Zt 0.3347 1.8893 1.1430
PSS-Sg 0.7503 0.8919 0.8737

Table 5  Loading conditions for Case 4


tuned PID and LL parameters by all seven techniques for
Case 2.
Loading condition Case 1 Speed deviation, phase plane and complex s-plane
Active power, P (p.u.) 0.2 response for Case 3 are shown in Fig. 13a–c, respectively.
Reactive power, Q (p.u.) 0.5 Similar pattern of results as in Case 1 and Case 2 are also
obtained in Case 3, in which PID-MO successfully gives the
shortest damping time, as well as the smallest oscillation
rate compared to other techniques. From the phase plane
3.5 s. This proved that all seven techniques have improved
response perspective, PID-MO stopped rotating after the
the angle stability very well. Among those techniques, PID-
second cycle, while the rest finished their rotation in almost
MO is significantly prominent method by the fastest damp-
6 cycles. From the result of eigenvalues region, it indicates
ing time, 1.7 s and the smallest oscillation rate compared to
that the PID-MO manages to move the eigenvalues posi-
other methods. On the other hand, systems installed with
tion to the most left and nearest to the real axis of com-
PSS-LL is more reliable to give better damping capability
plex s-plane. This implies that PID-MO is the best solution
compared to SVC-PI installed systems. Among all the seven
in solving the SMIB damping problem. Also, among the
approaches, PI-Sg takes the longest damping time. From
same devices, the application of FMO as objective function
MO-Zt-Sg aspect, the damping achievement of system with
is more reliable to tune parameter of devices compared to
PI-MO is more ahead than PI-Zt and PI-Sg, as well as PSS-
ξmin and σmax. Table 4 tabulated the value of tuned PID and
MO which also give better result as compared to the PSS-Zt
LL parameters for Case 3.
and PSS-Sg approach.
To compare the capability of the proposed optimization
The same results are also received for the phase plane
technique, the performance of SMIB that is installed with
response of Case 2 in Fig. 12b. The phase plan rotations of
conventional SVC-PID system (C-SVC) is compared with
all techniques are between 3 and 6 cycles. PID-MO shows
EP based SVC-PID system (EP-SVC), AIS based SVC-PID
the smallest cycle size, while other six methods provide
system (AIS-SVC) and PSO based SVC-PID system (PSO-
almost the same size of cycles.
SVC). The above comparison event is simulated in Case 4.
Figure 12c shown the sector of eigenvalues in complex
The loading condition for Case 4 is tabulated in Table 5.
s-plane for Case 2. All techniques have successfully shifted
Simulation is conducted based on the SMIB and SVC
the eigenvalues coordinates to the left region of origin in
parameters which can be obtained from [34]. In Case 4,
the complex s-plane. This indicate that all systems have
all value of KP, KI and KD parameters for C-SVC are ran-
achieved stable conditions. For PID-MO, their eigenvalues
domly selected. FMO is selected as the objective function
are located at the nearest to the real axis and the furthest to
in searching for optimal KP, KI and KD using EP, AIS and
the left side of complex s-plane. This place PID-MO as the
PSO methods.
best technique compared to the others. On the other hand,
The response of speed deviation and region of eigen-
some eigenvalues of PI-MO and PI-Zt are located near to
values location in complex s-plane for Case 4 is shown in
origin, verify that these two methods are bringing the SMIB
Fig. 14a, b, respectively.
system near to instability. Table 3 tabulated the value of
In Case 4, the PSO-SVC system shows the fastest
damping and the lowest oscillation compared to other

13
Journal of Electrical Engineering & Technology

Fig. 14  Speed response and eigenvalues for Case 4

three techniques as shown in Fig. 14a. It stops damping in


1.5 s. EP-SVC system is second place followed by AIS- Table 6  Correlation of AIS-SVC, EP-SVC, PSO-SVC and C-SVC for
SVC and C-SVC system. Figure 14b shows the region of Case 4
eigenvalues location in complex s-plane for Case 4. The Type KP KI KD
eigenvalues of PSO-SVC are positioned at the most left of
negative side, as well as at the nearest to the real axis of AIS-SVC 2.1323 − 5.0012 0.0112
the complex s-plane, compared to other three techniques. EP-SVC 2.0483 − 4.8066 0.0195
On the other hand, C-SVC system gives the worst result PSO-SVC 1.7312 − 4.3328 0.0243
as three of six eigenvalues located very near to right-hand C-SVC 2.500 − 5.0000 0.0100
side of complex s-plane. In overall result, PSO technique

13
Journal of Electrical Engineering & Technology

approach is capable to elevate the performance of SVC 7. Hannan MA, Islam NN, Mohamed A, Lipu MSH, Ker PJ,
system by optimizing the PID controllers. Rashid MM, Shareef H (2018) Artificial intelligent based damp-
ing controller optimization for the multi-machine power system:
The results of KP, KI and KD using EP, AIS and PSO a review. IEEE Access 6:39574–39594
methods for Case 4 are tabulated in Table 6. 8. Kumar A (2015) Power system stabilizers design for multima-
chine power systems using local measurements. IEEE Trans
Power Syst 31(3):2163–2171
9. Kamari NAM, Musirin I, Othman Z, Halim SA (2017) PSS
based angle stability improvement using whale optimization
6 Conclusion approach. Indones J Electr Eng Comput Sci 8(2):382–390
10. Tu G, Li Y, Xiang J, Ma J (2018) Distributed power system
This paper discussed the effectiveness of multi objec- stabilizer for multimachine power systems. IET Gener Transm
Distrib 13(5):603–612
tive function, FMO in tuning SVC-PID to improve angle 11. Islam NN, Hannan MA, Shareef H, Mohamed A (2017) An
stability in SMIB system. From the result, the introduc- application of backtracking search algorithm in designing power
tion of FMO manage to elevate the stability efficiency of system stabilizers for large multi-machine system. Neurocom-
power system. The application of minimum damping ratio, puting 237:175–184
12. Sung BC, Baek SM, Park JW (2014) Dynamic embedded opti-
ξmin as objective function of an optimization process will mization applied to power system stabilizers. J Electr Eng Tech-
move the eigenvalues region farther to left side of complex nol 9(2):390–398
s-plane, while the usage of maximum damping factor, σmax 13. Das S, Chatterjee D, Goswami SK (2018) Tuned-TSC based
will bring eigenvalues position towards the real axis of SVC for reactive power compensation and harmonic reduction
in unbalanced distribution system. IET Gener Transm Distrib
complex s-plane. Merger of ξmin and σmax index in certain 12(3):571–585
ratio as objective function will bring both advantages into 14. Wan Y, Murad MAA, Liu M, Milano F (2018) Voltage frequency
the selected optimization process. Based on three cases control using SVC devices coupled with voltage dependent loads.
that has been simulated, comparison of SVC-PID, SVC-PI IEEE Trans Power Syst 34(2):1589–1597
15. Kamari NAM, Musirin I, Hamid ZA, Ibrahim AA (2018) Optimal
and PSS-LL using various objective function revealed the tuning of SVC-PI controller using whale optimization algorithm
capability of SVC-PID in enhancing the stability perfor- for angle stability improvement. Indones J Electr Eng Comput Sci
mance, especially when collaborate with FMO as objective 12(2):612–619
function. From the optimization techniques point of view, 16. Farkoush SG, Kim CH, Jung HC, Lee S, Umpon NT, Rhee SB
(2017) Power factor improvement of distribution system with EV
PSO based optimization method is manage to calculate the chargers based on SMC method for SVC. J Electr Eng Technol
optimal parameters in enhancing the capability of damping 12(4):1340–1347
controller compared to EP and AIS. 17. Ugranli F, Karatepe E (2017) Coordinated TCSC allocation and
network reinforcements planning with wind power. IEEE Trans
Acknowledgements This work was supported by Universiti Kebang- Sustain Energy 8(4):1694–1705
saan Malaysia under code GGPM-2018-055. 18. Islam NN, Hannan MA, Mohamed A, Shareef H (2016) Improved
power system stability using backtracking search algorithm for
coordination design of PSS and TCSC damping controller. PLoS
One 11(1):e0146277
References 19. Bizzarri F, Brambilla A, Milano F (2017) Analytic and numerical
study of TCSC devices: unveiling the crucial role of phase-locked
1. Song Y, Hill DJ, Liu T (2017) Network-based analysis of small- loops. IEEE Trans Circuits Syst I Regul Pap 65(6):1840–1849
disturbance angle stability of power systems. IEEE Trans Con- 20. Wang L, Chang CH, Kuan BL, Prokhorov AV (2017) Stability
trol Netw Syst 5(3):901–912 improvement of a two-area power system connected with an inte-
2. Kamaruzzaman ZA, Mohamed A, Kamari NAM (2018) Effect grated onshore and offshore wind farm using a STATCOM. IEEE
of grid-connected photovoltaic generator on dynamic voltage Trans Ind Appl 53(2):867–877
stability in power system. J Eng 30(2):289–296 21. Shokri A, Shareef H, Mohamed A, Zayandehroodi H, Farhoodnea
3. Chitara D, Niazi KR, Swarnkar A, Gupta N (2018) Cuckoo M (2013) Effect of network parameters on the injection current
search optimization algorithm for designing of a multimachine requirement of D-STATCOM designed to mitigate voltage sag.
power system stabilizer. IEEE Trans Ind Appl 54(4):3056–3065 Res J Appl Sci Eng Technol 5(1):161–166
4. Kamari NAM, Musirin I, Hamid ZA, Rahim MNA (2012) 22. Varma RK, Maleki H (2019) PV solar system control as STAT-
Optimal design of SVC-PI controller for damping improvement COM (PV-STATCOM) for power oscillation damping. IEEE
using new computational intelligence approach. J Theor Appl Trans Sustain Energy 10(4):1793–1803
Inf Technol 42(2):271–280 23. Movahedi A, Niasar AH, Gharehpetian GB (2019) LVRT improve-
5. Sun Z, Cai G, Yang D, Liu C, Wang B, Wang L (2018) A ment and transient stability enhancement of power systems based
method for the evaluation of generator damping during low- on renewable energy resources using the coordination of SSSC
frequency oscillations. IEEE Trans Power Syst 34(1):109–119 and PSSs controllers. IET Renew Power Gener 13(11):1849–1860
6. Islam NN, Hannan MA, Mohamed A, Shareef H (2016) Damp- 24. Mahto T, Mukherjee V (2015) Frequency stabilisation of a hybrid
ing power system oscillation using elitist differential search two-area power system by a novel quasi-oppositional harmony
algorithm in multi machine power system. J Theor Appl Inf search algorithm. IET Gener Transm Distrib 9(15):2167–2179
Technol 93(1):41–47 25. Rajaram T, Reddy JM, Xu Y (2016) Kalman filter based detection
and mitigation of subsynchronous resonance with SSSC. IEEE
Trans Power Syst 32(2):1400–1409

13
Journal of Electrical Engineering & Technology

26. Chen J, Tao J, Wang C, Wei P, Liu J, Li Q, Zhou Q (2019) Control DC motor speed control with back EMF detection. J Electr Eng
strategy of UPFC based on power transfer distribution factor. J Technol 11(3):715–723
Eng 2019(16):1897–1899 46. Kamari NAM, Musirin I, Hamid ZA, Zaman MHM (2018) Oscil-
27. Sen D, Acharjee P (2019) Optimal line flows based on voltage pro- latory stability prediction using PSO based synchronizing and
file, power loss, cost and conductor temperature using coordinated damping torque coefficients. Bull Electr Eng Inform 7(3):331–344
controlled UPFC. IET Gener Transm Distrib 13(7):1132–1144 47. Nasir SCM, Mansor MH, Musirin I, Othman MM, Kuan TM,
28. Ebeed M, Kamel S, Yu J, Jurado F (2019) Development of UPFC Kamil K, Abdullah MN (2019) Multistage artificial immune sys-
operating constraints enforcement approach for power flow con- tem for static var compensator planning. Indones J Electr Eng
trol. IET Gener Transm Distrib 13(20):4579–4591 Comput Sci 14(1):346–352
29. Khiabani AG, Babazadeh R (2016) Design of robust fractional- 48. Alonso FR, Oliveira DQ, Zambroni de Souza AC (2014) Arti-
order lead-lag controller for uncertain systems. IET Control The- ficial immune systems optimization approach for multiobjec-
ory Appl 10(18):2447–2455 tive distribution system reconfiguration. IEEE Trans Power Syst
30. Abdulkhader HK, Jacob J, Mathew AT (2018) Fractional-order 30(2):840–847
lead-lag compensator-based multi-band power system stabiliser 49. Dudek G (2016) Artificial immune system with local feature selec-
design using a hybrid dynamic GA-PSO algorithm. IET Gener tion for short-term load forecasting. IEEE Trans Evol Comput
Transm Distrib 12(13):3248–3260 21(1):116–130
31. Lu CF, Hsu CH, Juang CF (2012) Coordinated control of flexible 50. Hasanien HM (2018) Whale optimisation algorithm for auto-
ac transmission system devices using an evolutionary fuzzy lead- matic generation control of interconnected modern power sys-
lag controller with advanced continuous ant colony optimization. tems including renewable energy sources. IET Gener Transm
IEEE Trans Power Syst 28(1):385–392 Distrib 12(3):607–614
32. Amiri MS, Ramli R, Ibrahim MF (2019) Hybrid design of PID 51. Prasad D, Mukherjee A, Shankar G, Mukherjee V (2017) Appli-
controller for four dof lower limb exoskeleton. Appl Math Model cation of chaotic whale optimisation algorithm for transient sta-
72:17–27 bility constrained optimal power flow. IET Sci Meas Technol
33. Chaiyatham T, Ngamroo I (2014) Improvement of power system 11(8):1002–1013
transient stability by PV farm with fuzzy gain scheduling of PID 52. Jang SH, Roh JH, Kim W, Sherpa T, Kim JH, Park JB (2011)
controller. IEEE Syst J 11(3):1684–1691 A novel binary ant colony optimization: application to the unit
34. Kamari NAM, Musirin I, Othman MM (2017) IPSO based SVC- commitment problem of power systems. J Electr Eng Technol
PID for angle stability enhancement. Int J Simul Syst Sci Technol 6(2):174–181
17(41):20.1–20.7 53. Yin Z, Du C, Liu J, Sun X, Zhong Y (2017) Research on auto-
35. Yin XX, Lin YG, Li W, Liu HW, Gu YJ (2015) Fuzzy-logic slid- disturbance-rejection control of induction motors based on an
ing-mode control strategy for extracting maximum wind power. ant colony optimization algorithm. IEEE Trans Ind Electron
IEEE Trans Energy Convers 30(4):1267–1278 65(4):3077–3094
36. Torres SP, Peralta WH, Castro CA (2007) Power system load- 54. Suhane P, Rangnekar S, Mittal A, Khare A (2016) Sizing and per-
ing margin estimation using a neuro-fuzzy approach. IEEE Trans formance analysis of standalone wind-photovoltaic based hybrid
Power Syst 22(4):1955–1964 energy system using ant colony optimisation. IET Renew Power
37. Hannan MA, Ghani ZA, Hoque MM, Ker PJ, Hussain A, Gener 10(7):964–972
Mohamed A (2019) Fuzzy logic inverter controller in photovol-
taic applications: issues and recommendations. IEEE Access Publisher’s Note Springer Nature remains neutral with regard to
7:24934–24955 jurisdictional claims in published maps and institutional affiliations.
38. Kim MK (2015) A new approach to short-term price forecast strat-
egy with an artificial neural network approach: application to the
nord pool. J Electr Eng Technol 10(4):1480–1491
39. Subramani C, Jimoh AA, Kiran SH, Dash SS (2016) Artificial Dr. Nor Azwan Mohamed Kamari received Bachelor in Electrical
neural network based voltage stability analysis in power system. and Electronic Engineering from Meiji Univeristy, Japan, M.Sc. in
In proceedings of international conference on circuit, power and Electrical and Electronic Engineering from Ehime University, Japan
computing technologies (ICCPCT), Nagercoil, India and Ph.D. in Electrical Engineering from Universiti Teknologi Mara,
40. Martin F, Aguado JA (2003) Wavelet-based ANN approach Malaysia in 2000, 2004 and 2016, respectively. He has been an acade-
for transmission line protection. IEEE Trans Power Deliv mician for over 10 years at several universities before joining Universiti
18(4):1572–1574 Kebangsaan Malaysia (UKM) in 2017. His areas of research interests
41. Wang L, Zhang Q, Zhou A, Gong M, Jiao L (2015) Constrained are power engineering, energy management and artificial intelligence.
subproblems in a decomposition-based multiobjective evolution- He has been the project leader for several research projects with fund-
ary algorithm. IEEE Trans Evol Comput 20(3):475–480 ing from the Ministry of Education Malaysia and UKM. He oversees
42. Kamari NAM, Musirin I, Othman MM (2011) Computational several postgraduate, local and international students in the field of
intelligence technique-based pi controller using SVC. In: Pro- power engineering. He also is a reviewer for various local and inter-
ceeding of IEEE power engineering and automation conference, national journals.
Wuhan, China
43. Tsai MS, Hsu FY (2009) Application of grey correlation analysis Prof. Ir. Dr. Ismail Musirin obtained Bachelor of Electrical Engi-
in evolutionary programming for distribution system feeder recon- neering (Hons) in 1990 from Universiti Teknologi Malaysia, M.Sc.
figuration. IEEE Trans Power Syst 25(2):1126–1133 Pulsed Power Technology in 1992 from University of Strathclyde,
44. Anbarasi S, Muralidharan S (2017) Hybrid BFPSO approach United Kingdom and Ph.D. in Electrical Engineering from Univer-
for effective tuning of PID controller for load frequency Control siti Teknologi MARA (UiTM), Malaysia in 2005. He is currently a
application in an interconnected power system. J Electr Eng Tech- Professor of Power System at the Faculty of Electrical Engineering,
nol 12(3):1027–1037 UiTM and headed the Power System Operation (POSC) Computational
45. Kiree C, Kumpanya D, Tunyasrirut S, Puangdownreong D Intelligence Research Group. He has authored 2 books published, over
(2016) PSO-based optimal PI(D) controller design for brushless 300 papers in international indexed journals and conferences. He is
also an international journal reviewer for IEEE Transactions, Elsevier

13
Journal of Electrical Engineering & Technology

Science, WSEAS, John Wiley, IET and some other publishers. He has System Society (ARTIST) and member of International Association of
chaired more than 20 international conference since 2007. To date, Engineers (IAENG), Hong Kong.
he has delivered keynote speeches at Cambridge University, United
Kingdom, Dubai, Korea and Malaysia. He has also been given oppor- Dr. Ahmad Asrul Ibrahim received B.Eng. and M.Sc. degrees from Uni-
tunity to evaluate research grants at the national and international lev- versiti Kebangsaan Malaysia, Malaysia and Ph.D. degree from Durham
els. His research interest includes artificial intelligence, optimization University, United Kingdom in 2008, 2012 and 2018, respectively. He
techniques, power system analysis, renewable energy, distributed gen- has authored and co-authored more than 30 technical papers published
eration and power system stability. He is a professional engineer and in the international journals, national and international conferences.
a senior member of International Association of Computer Science His research interests include distribution system automation, artificial
and Information Technology (IACSIT), member of Artificial Immune intelligence and power quality assessment.

13

You might also like