Journal of Electrical Engineering

www.jee.ro

LOAD COMPENSATION AND VOLTAGE REGULATION IN THREE-PHASE FOUR-

WIRE DISTRIBUTION SYSTEM USING PHOTOVOLTAIC SUPPORTED
CUSTOM POWER DEVICE

Dr. C.K. Sundarabalan, Dr. K. Selvi

Assistant Professor Associate Professor
EEE Department EEE Department
P.R.Engineering College Thiagarajar College of Engineering
Thanjavur, Tamilnadu, India Madurai, Tamilnadu,India
[email protected] [email protected]

Abstract: The necessity of good quality power is the mitigating current harmonics, load balancing, voltage
obligatory one for todays power system due to the rapid regulation, neutral current compensation etc [2,3]. SSTS
growth of a nonlinear load, unbalanced load and is one of a network reconfiguring type CPD and it is able
sensitive load utilization in three-phase four-wire to protect the load from voltage disturbances by
distribution system. In this paper, the photovoltaic (PV) transferring from the preferred feeder to the alternate
supported Distribution Static Compensator (DSTATCOM) feeder [4]. UPQC is another compensating type CPD,
is used to mitigate unbalanced source current, current which is the combination of shunt and series active filter
harmonics and neutral current. In addition, the thyristor- with a common DC capacitor and it is able to compensate
based solid state transfer switch (SSTS) is used in the both voltage and current related problems simultaneously
proposed system to provide voltage regulation by [5]. Four-leg VSC based DSTATCOM eliminate the
transferring from the preferred source to the alternate excessive neutral current without the help of a transformer
source in case of a fault in the preferred source. [6]. In addition, this four-leg topology is most suitable for
Synchronous reference frame (SRF) theory is used to compensation of high neutral currents [7,8].
extract the reference current signals for the proposed It has been observed from literature survey, the
DSTATCOM. The performance investigation of Adaptive dynamic performance of the CPD is mainly dependent on
Neuro Fuzzy Inference System (ANFIS) based DC voltage its control strategy. It is renowned that the several
regulation is compared with the Proportional Integral investigations have been carried out for deriving the
(PI) controller under nonlinear and unbalanced linear reference current signals for the DSTATCOM. The
load conditions. The effective voltage sag/swell detection instantaneous reactive power theory (pq theory) [9], SRF
is proposed to extract the transfer signal for SSTS. This theory [10], Icos theory [11], sliding mode control [12]
paper shows that, the SSTS along with PV supported four- etc. have been proposed to control the DSTATCOM. Due
leg voltage source converter (VSC) based DSTATCOM to lower computation, the SRF theory is used in this work
topology simultaneously compensate both the current and for deriving the reference current signals.
voltage disturbances in three-phase four-wire distribution Artificial Intelligence has been realized as a suitable
system. The effectiveness of the proposed system is tool for the analysis of power quality disturbances
analyzed in Matlab/Simulink. [13,14]. A DC voltage regulator is used to maintain the
DC-link voltage of DSTATCOM at constant and equal to
Key words: ANFIS controller, PI controller, the specified reference value. Artificial intelligence based
DSTATCOM, power quality, Voltage Sag DC-link voltage controller gives better performance than
PI controller [15]. The ANFIS combines both the learning
1. Introduction abilities of a neural network and reasoning abilities of
In present day electric power distribution systems, the fuzzy logic [16]. In this work, the ANFIS based DC
quality of power is highly distorted due to extensive voltage regulation is proposed. Renewable energy placed
application of power electronic devices and other a significant role in power quality improvement. As a
distorted loads [1]. In recent years, there has been a matter of fact, the PV generation is considered as an
growing interest in custom power device (CPD) for power important source in present day electric power system
quality enhancement in distribution system. There are [17,18]. Recently researchers have focused on the
several types of CPD but the DSTATCOM and SSTS are distributed generation (DG) supported CPD for power
considered in this work. DSTATCOM is a shunt quality improvement in distribution system [11,15,19].
connected compensating type CPD and its capable of

1
 Journal of Electrical Engineering
www.jee.ro

In this paper, the PV supported four-leg to the alternate feeder. In order to eliminate reactive
DSTATCOM along with SSTS is modelled to mitigate current and harmonic current in the source side the
current and voltage related power quality problems in proposed shunt connected DSTATCOM injects the
three-phase four-wire distribution system. The PV array appropriate converter current. Further, the fourth leg of
or battery is used to maintain the DC-link voltage of the the DSTATCOM provides neutral current elimination in
four-leg DSTATCOM. The proposed system compensates the source side. The PV array or battery is used to
neutral current, harmonic current, voltage sag, voltage maintain the DC capacitor voltage through the DC-DC
swell and load balancing. Further, the performance of boost converter. The ripple filter is connected in shunt
ANFIS based DC-link voltage regulation is investigated with the load terminals to eliminate switching harmonics
with the performance of PI controller under nonlinear and produced by the VSC. All the parameters used in the
unbalanced linear load conditions. Matlab/Simulink planned system are given in the Appendix.
results are provided to validate the dynamic performance
of the proposed system. 3. Photovoltaic

2. Configuration The PV systems are renowned as a significant

alternative among the renewable energy sources that will
The Table 1 shows the difference between the secure the availability of energy in future [17, 23]. The
proposed topology with the existing SRF based PV array composed of several PV modules
DSTATCOM topology. Several work in this area mainly
focused on load compensation alone. In [2] and [20] both qVPV / N s I PV Rse / N p
voltage regulation and load balancing are done but KTc A N V / N I R
simultaneous compensation is not fulfilled. The proposed I PV N P I ph N p I o e 1 P PV

s PV se
(1)
Rsh

work focuses on the simultaneous compensation in
distribution system using SSTS with DSTATCOM.
Where, IPV is the PV module current (A), VPV is the
Table 1 Comparison of Existing SRF based DSTATCOM terminal voltage (V), Iph is the light generated current in a
topology with the proposed system. PV module (A), IO is the PV panel reverse saturation
current (A), K is Boltzmann constant (1.381023 J/K),
Existing topology TC is the junction temperature (K), q is the electron
DSTATCOM Proposed
Topology system charge (1.609 1019 C), NP is the number of cells in
[2] [19] [20] [21] [22] parallel, NS is the number of cells in series, Rse is the
series resistance () and Rsh is the parallel resistance ().
Load
The SOLKAR-36 W PV module proposed in this work
balancing
has been developed based on the model presented and
Voltage
validated in [24]. The determination of Rse and Rsh is
regulation
carried out using [25]. The table 2 summarizes the
DG
parameters of the PV module.
supported
DSTATCOM Table 2 Electrical parameters of the PV module.
Four-leg

VSC Parameter Value
ANFIS Voc 21 V

controller Isc 2.5 A
SSTS Vmpp 16.56 V
Impp 2.25 A
The detailed configuration of the proposed three- Rse 0.455
phase four-wire system is shown in Fig. 1. The power Rsh 174.4156
supply to the nonlinear or unbalanced load is supplied by Ideality factor 1.6
the two sources namely preferred source and alternate No. of cells in series 36
source. During normal operations, the preferred source No. of cells in parallel 1
supplies power to the load through the preferred feeder.
When a voltage disturbance occurs in the preferred feeder
or preferred source, the SSTS quickly transfers the supply

2
 Journal of Electrical Engineering
www.jee.ro

Preferred source Z2 AP1 Z1 Alternate source

IPa PP1 IAa
A A
PN1 P AN1
IPb P2 AP2 IAb
N B B N
IPc PN2 PP3 AN2 AP3 IAc
C C
PN3 AN3 A
IPn PP4 P4 IAn

PN4 AN4 SSTS

ISa ISb ISc ISn

S1 Nonlinear Load
ICa iLa

ICb iLb

ICc iLc

ICn iLn

S2 Unbalanced Load
Ripple filter
A

L B

DC-DC
Boost converter
+ +
C DC
VDC VPV PV
DC
-
-
+ -
Four-leg VSC Battery

Fig. 1. Schematics of proposed methodology.

3
 Journal of Electrical Engineering
www.jee.ro

The required input voltage of the DC-DC converter DSTATCOM during night-time and lower irradiation
can be obtained from (1-0.55) 700=315 V. Where 0.55 condition.
and 700 are the duty ratio and output voltage of the DC-
DC converter respectively. The design of DC-DC 4. SSTS
converter is carried out using [26]. The rated operating
point of the proposed PV module is 2.25 A and 16.56 V. The SSTS can be used very effectively to protect
The number of PV modules required is calculated by sensitive loads against voltage sag, swell and other
using Eqs. (2) and (3). electrical disturbances [1,27,28]. The SSTS is the
combination of two thyristor block [4]. Each block
VPV 315 composed of four thyristor modules corresponding to the
PVseries 19 (2) three-phase four-wire system. Each thyristor module
VPV 1 16.56 composed of two sets of thyristor switches connected in
PPV 6 kW opposed directions. Preferred feeder voltage is
PV parallel 9 (3) continuously monitored by the proposed control logic.
PV series 0.36 kW 19 0.36 kW
When the preferred source has an appropriate voltage,
The effect of irradiance and temperature on the control logic turns on thyristors on the preferred feeder
current-voltage (I-V) characteristics of PV module is side. If any voltage sag or swell is detected from the
shown in Fig. 3. It is observed that, the higher irradiance preferred side voltage then alternate feeder transfer is
with lower temperature generates maximum PV power. initiated by removing gating pulses from the thyristors on
the preferred feeder switch and firing thyristors on the
alternate feeder side.

4.1. Voltage Disturbances Detection Logic

Vd Vd2
VPa x
x2
VPb abc - dq +
Transform Vq Vq2 +
VPc x 2

Vdq
sin cos LPF
-
Phase +
Locked u
Absolute
Loop 1
Reference
Sag /Swell
Detection
signal Hysteresis

Fig. 4. Block diagram of proposed voltage sag/swell

detection.
Fig. 3. Effect of irradiance and temperature on the I-V
characteristics of PV module. The block diagram of the proposed voltage sag/swell
detection method is shown in Fig. 4. The three phase
The proposed PV array composed of (199) modules voltage (Vpa, Vpb and Vpc) from preferred source are
with the power rating of 6 kW. The DC-DC converter is transformed into (Vd and Vq).
used to boost up the incoming PV voltage from 315 V to
Vd VPa sin(t ) VPb sint 2 / 3 VPc sint 2 / 3
700 V DC. The Battery is used to support the 2
(4)
3

4
 Journal of Electrical Engineering
www.jee.ro

I loss(n) I loss( n1) K pd (Vde(n) Vde(n1) ) K id Vde(n) (11)

Vq VPa cos(t ) VPb cos t 2 / 3 VPc cost 2 / 3 (5)
2
3
Where Vde(n) is the error between reference and sensed DC
Vdq Vd2 Vq2 (6) voltage at the n th sampling instant. Kpd and Kid are the
The obtained Vdq signal from Eq. (6) is compared with the proportional and integral gains. Therefore the reference
reference value (1 pu) and filtered using LPF. The source current is,
absolute value of the resulting variable is subjected to a
hysteresis comparator and the output of this comparator I d* I d1 I loss (12)
generates the sag/swell detection signal. This signal is 0
when a sag/swell occurs, otherwise it is 1.
Vdc ref Isa Isb Isc
5. DSTATCOM
Calculate I a*
The DSTATCOM composed of four-leg VSC PI PL reference Ib *
connected in shunt with the DC capacitor [2]. When
Vd -+ controller + currents I c*
DSTATCOM connected in the load terminals, it will c
Id
inject the converter current in order to eliminate harmonic ILa ILabc LPF + Idq
contents, DC offset in load and cancels the effect of Iddc Id *
ILb to Iq to
unbalanced loads in the distribution system. Moreover ILc Iqdc
Idq0 LPF + ISabc
DSTATCOM can perform voltage regulation while it is Iq *
connected to a distribution bus [1,29]. In this paper, the cos, + I
DSTATCOM is connected in shunt with load terminals, sin PLL cos, qr

so it is capable of performing load compensation only. To sin

overcome the above shortcoming, SSTS is used for Amplitude - PI
voltage regulation. The performance of the DSTATCOM + controller
mainly depends on the control algorithms, switching V
scheme and design of circuit components [30]. s Vs *
Vsa Vs Vsc
5.1 DSTATCOM controller b
Fig. 5. Generation of reference current using PI controller
The extraction of reference currents using PI based SRF theory.
controller based SRF theory is shown in Fig. 5. The load
currents from a-b-c co-ordinates (ILa, ILb, ILc) are Similarly in quadrature component, average value of
converted into d-q-0 co-ordinates (Id, Iq, I0 ). DC component is referred as Iq1 and oscillating quantity
as Iq2.
Id
2
I La sin(t ) I Lb sint 2 / 3 I Lc sint 2 / 3 (7) I q I q1 I q 2 (13)
3

I q I La cos(t ) I Lb cost 2 / 3 I Lc cost 2 / 3 (8)

2 I qr(n) I qr( n1) K pq (Vte( n) Vte(n1) ) K iqVte(n) (14)
3
I q* I q1 I qr (15)
I 0 I La I Lb I Lc
1
(9) The reference source currents in the abc frame are
3
obtained by reverse transformation of the current vector.
Each current component has an average value of DC
component referred as Id1 and an oscillating value or ac I a I d* sin t I q cos t (16)
component as Id2. I b I d* sint 2 / 3 I q cos t 2 / 3 (17)

I d I d1 I d 2 (10) I
c I d* sint 2 / 3 I q cos t 2 / 3 (18)
The sensed source currents (Isa, Isb, Isc ) are compared with
The output of PI controller at the DC bus voltage of the generated reference source currents (Ia*, Ib*, Ic*) and
DSTATCOM is considered as the current (iloss) for the resulted current is fed to the hysteresis controller to
meeting its losses. generate gate pulses. The parameters of DC-link voltage
PI controller are same as [15].

5
 Journal of Electrical Engineering
www.jee.ro

5.2. Generation of Neutral current reference signal.

The extraction of reference current signal for fourth-
leg of VSC in DSTATCOM is obtained by using Eq. (19). START

i n i La i Lb i Lc (19)
Load Training Data and
Checking Data
The switching signals for the fourth-leg of VSC in
DSTATCOM are generated from the error signal by
comparing the load current and reference [2]. The Generate Fuzzy Inference System (FIS)
extracted signal is fed to hysteresis controller to generate using generalized bell-shaped membership function
switching pulses for the fourth-leg of VSC in
DSTATCOM.

5.3 ANFIS controller based DC voltage regulation in Select Hybrid

four-leg DSTATCOM Learning Method

The ANFIS utilize the learning mechanism of neural

network for determining the membership function
Set the number of Epochs =40 and
parameters of the fuzzy logic
[16,31]. Error tolerance=0
L4
L1 x1 x2 Start Training
L2 L3
A1
x1 w1 w1

....

N1 1 L5
.

A7 1 w1 f1



.

Input-1 NO Training
y completed
B1 Output YES
x2
..

N4 49 w49 f 49
w49 Load checking data and start testing
B7 49 w49 9

Premise x1 x2
Input-2 Consequent
NO Testing
Fig. 6. ANFIS architecture. completed
The data collected from input and output of the PI
YES
controller are used to train the ANFIS controller. The first
input (x1) and the second input (x2) is the error and change
Save FIS
in error between the actual and reference DC voltage
respectively. The proposed ANFIS controller has five
layers, two inputs (x1 and x2) and one output(y) is shown
in Fig. 6. The node functions of five layers displayed in END
the ANFIS architecture are enlightened as follows [16].
The node presented in layer-1 is symbolized by a square.
Where x1 is the input to the node i, and Ai is the linguistic Fig. 7. Flowchart of ANFIS training and testing in
label (A1, A2, A3, A4, A5, A6 and A7) linked with this Matlab.
node function. The detailed procedure to generate ANFIS
controller in Matlab is shown in Fig. 7.

6
L
 Journal of Electrical Engineering
www.jee.ro

In this work, seven generalized bell-shaped membership rj, sj, tj = consequent parameter set which are
functions are used and the corresponding node equations determined during training
are given below. Ai, Bi = fuzzy membership function
The square node is an adaptive node and its parameters
Oi1 Ai x1 (20) are changed during training but the parameters presented
The parameter in layer-1 is referred as premise in circle nodes are fixed. The initial and final membership
parameters. The node presented in layer-2 is symbolized functions of Input-1 and Input-2 of ANFIS controller are
by a circle and labeled as . shown in Fig. 8.

Ai x1
1
2 bi
(21)
x c
1 1 i
ai
These layers multiply the incoming signals and forward it
to the layer-3.
w j Ai x1 Bi x2 (22)
The output from each node in layer-2 represents the firing
strength of a rule. Each node presented in layer-3 is (a) Initial generalized bell-shaped membership function of
symbolized by a circle and labeled as N. The output of input-1 (x1).
this layer calculates the normalized firing strength of
every rule.
wj
w j 49 (23)
w
k 1
k

Each node presented in layer-4 is symbolized by a square.

The parameter in layer-4 is referred as consequent
parameters.

O 4j w j f j w j r j x1 s j x2 t j (24)
Rule base (b) Initial generalized bell-shaped membership function
of input-2 (x2).
If x1is A1and x2 is B1 then f1 r1 x1 s1 x2 t1
If x1is A1and x2 is B2 then f 2 r2 x1 s 2 x2 t 2
.
.
If x1is A7 and x2 is B7 then f 49 r49 x1 s 49 x2 t 49
Single node is presented in layer-5 which is symbolized
by a circle and labeled as . This layer calculates the
output (y) by summing up of all incoming signals.

w x r w x s w t
49 49
y w
j 1
j fj
j 1
j 1 j j 2 j j j (25)
(c) Final generalized bell-shaped membership function of
Where, input-1(x1).
i =1, 2 7 and j=1, 2 49.
Oi1 = output of the ith node in layer1
ai, bi, ci = Parameters of generalized bell-shaped
membership function
Oj4 = output of the ith node in layer4
w j = Layer-3 output

7
 Journal of Electrical Engineering
www.jee.ro

(d) Final generalized bell-shaped membership function of Fig. 10. Surface view of ANFIS controller.
input-2 (x2).
Table 3 ANFIS Parameters used for Training.
Fig. 8. Initial and final generalized bell-shaped
membership function of input-1 and input-2. Parameter of ANFIS value
Number of Epoch 40
The proposed ANFIS uses a hybrid learning
Number of nodes: 131
algorithm which is the combination of the least-squares
Number of linear parameters: 49
method and the backpropagation gradient descent method.
Number of nonlinear parameters: 42
In hybrid learning every iteration composed of the
Total number of parameters: 91
forward pass and backward pass. During forward pass, the
Number of training data pairs: 25001
premise parameters are fixed and the consequent
Number of checking data pairs: 25000
parameters are adjusted by the least square method, while
Number of fuzzy rules: 49
in backward pass, the error signals propagate backwards
here the consequent parameters are fixed and the premise Isa Isb Isc
parameters are adjusted by the gradient descent method Vdc ref
[16]. Calculate I a*
ANFIS PL reference Ib *
Vd -+ controller + currents I c*
c
Id
ILa ILabc LPF + Idq
Iddc Id *
ILb to Iq to
ILc Iqdc
Idq0 LPF + ISabc
Iq *
cos, +
sin PLL cos,
sin Iqr
Amplitude - PI
+ controller
V

Fig. 9. Training error versus epochs plot.

s Vs *
Vsa Vs Vsc
b
The ANFIS used here contains 49 rules, with 7
Fig. 11. Generation of reference current using ANFIS
generalized bell-shaped membership functions being
controller based SRF theory.
assigned to each input variable. 25001 training data was
used for training of ANFIS and 25000 checking data was The extraction of reference current signals with
used for verifying the identified ANFIS. Fig. 9 shows the ANFIS controller based SRF theory is shown in Fig. 11.
training error values for each epoch. The three The ANFIS controller controls the small amount of active
dimensional surface view of generated ANFIS controller current by comparing the actual DC-link voltage of
is shown in Fig. 10. The detailed information about the DSTATCOM with the reference voltage and the output of
proposed ANFIS controller is given in Table 3. this controller is the corresponding power flow needed to
maintain the DC-link voltage. Hence this power flow is
added as a part of reference for current controller which

8
 Journal of Electrical Engineering
www.jee.ro

controls the inverter to provide the required compensation OFF. The three-phase voltage swell is created in the
current and maintain the DC-link voltage of the preferred feeder from 0.35 s to 0.45 s. It is observed that
DSTATCOM. 50% three-phase voltage swell is occurred in the VP. Due
to effective control strategy of the SSTS, the preferred
6. Results and Discussions feeder switches are turned OFF and the alternate feeder
switches turns ON at 0.354 s. So that, the point of
The performance of the proposed SRF based common coupling voltages (VS) will not be affected by
DSTATCOM is analyzed with both nonlinear and the voltage swell disturbance. After the completion of the
unbalanced linear load under voltage sag and swell voltage swell duration, the preferred feeder again
conditions. connects to the load and the alternate feeder switches are
turned OFF. However the proposed control strategy
6.1. Performance of the DSTATCOM with the detects voltage sag very quickly than voltage swell.
nonlinear load condition

The load balancing, voltage regulation and current

harmonic compensation performance of the DSTATCOM
with SSTS under nonlinear load condition is shown from
Fig. 12 to Fig. 17. The SSTS transfer signal for both
preferred feeder and alternate feeder under voltage sag
and swell condition is shown in Fig. 12. The nine subplots
of Figs. 13 and 14 illustrate, preferred source voltages
(VP), alternate source voltages (VA), point of common
coupling voltages (VS), preferred source currents(IP),
alternate source currents (IA), load currents (IL), converter
currents (IC), source currents (IS) and DC-link voltage
(VDC) under nonlinear load condition.

Fig. 13. Voltage regulation and load compensation with

Fig. 12. SSTS transfer signal under voltage sag and swell
PI controller under nonlinear load with voltage sag and
conditions.
swell conditions.
Figs.13 and 14 illustrate the performance of the PI
It is observed from Figs. 13 and 14, due to the
and ANFIS based DSTATCOM with SSTS under
presence of nonlinear load in the distribution system, the
nonlinear load with voltage sag and swell conditions. The
load current (IL) waveform is highly distorted. The PI
three-phase to ground fault is applied in the preferred
controller and ANFIS controller based DSTATCOM
feeder from 0.05 s to 0.15 s. It is observed from Fig. 13
compensate the harmonic current by injecting the required
and 15, 65% three-phase voltage sag is occurred in the VP.
converter currents (IC). So that, the source currents (IS) are
Due to effective control strategy of the SSTS, the
made sinusoidal, balanced and in-phase with the source
preferred feeder switches are turned OFF and the alternate
voltage.
feeder switches turns ON at 0.052 s. So that, the point of
common coupling voltages (VS) is not affected by the
voltage sag disturbance. After the completion of the
voltage sag duration, the preferred feeder again connects
to the load and the alternate feeder switches are turned

9
 Journal of Electrical Engineering
www.jee.ro

(b) Phase-b

(c) Phase-c

Fig. 14. Voltage regulation and load compensation with Fig. 15. Harmonic spectrum of source current before
ANFIS controller under nonlinear load with voltage sag compensation.
and swell conditions.

It is important to note that, the three-phase nonlinear

load is changed to two-phase load from 0.2 s to 0.3 s. Due
to the effective compensation of DSTATCOM the source
currents are still sinusoidal and balanced even when the
system is operated in two-phase load. Both the controllers
have good performance in load balancing but the ANFIS
controller has better DC voltage (VDC) regulation than the
PI controller under nonlinear load condition.

(a) Phase-a

(a) Phase-a

(b) Phase-b

10
 Journal of Electrical Engineering
www.jee.ro

Fig. 15 illustrates the performance of total harmonic

distortion (THD) before compensation in phase-a, phase-b
and phase-c of source current are 25.73%, 26.04% and
24.82% respectively. It is observed from Fig. 16, after
compensation with PI controller the THD of source
current in phase-a, phase-b and phase-c are reduced to
3.32%, 3.38% and 3.27% respectively. In case of ANFIS
controller the THD is reduced to 2.96%, 3.32% and
3.12% is shown in Fig. 17. It is noted that ANFIS
(c) Phase-c controller has better harmonic compensation than PI
Fig. 16. Harmonic spectrum of source current after controller.
compensation with PI controller.
Table 4 gives the comparison of the individual
harmonic level of source current between the proposed PI
controller and ANFIS controller. From this comparison, it
is clearly shown that the harmonics presented in the
source current are eliminated successfully. It is observed
that the THD of Is obtained from both PI controller and
ANFIS controller fulfill the requirement of IEEE-519
standard. However the ANFIS controller offers better
harmonic compensation than PI controller.

Table 4 Harmonic Contents of the source current.

(a) Phase-a Source current Harmonic Content (% of fundamental)

Before compensation After compensation
H
PI controller ANFIS controller
a b c a b c a b c
3 0.6 0.5 0.6 0.3 0.7 0.8 0.3 0.0 0.1
5 19.2 19.3 18.2 1.5 1.3 1.1 1.3 1.47 1.1
7 12.4 13.0 12.8 0.9 1.0 1.2 0.8 1.1 0.9
9 0.4 0.5 0.7 0.2 0.1 0.2 0.0 0.1 0.1
11 7.4 7.4 6.9 0.5 0.6 0.8 0.6 0.8 0.9
13 5.5 5.4 5.6 0.6 0.5 0.8 0.8 0.7 0.7
15 0.3 0.2 0.4 0.1 0.3 0.3 0.1 0.0 0.1
17 4.0 3.9 3.2 0.8 0.3 0.5 0.3 0.5 0.6
(b) Phase-b 19 3.1 3.2 3.2 0.3 0.5 0.6 0.2 0.0 0.3
THD 25.7 26.0 24.8 3.3 3.3 3.2 2.9 3.3 3.1
H -Harmonic order

6.2. Performance of the DSTATCOM with the

unbalanced linear load conditions

The performance of the DSTATCOM with SSTS

under unbalanced linear load condition is shown from
Fig. 18 to Fig. 20. The nine subplots of Fig. 18 and Fig.
19 illustrate, preferred source voltages (VP), alternate
source voltages (VA), point of common coupling voltages
(c) Phase-c (VS), preferred currents (IP), alternate currents (IA), load
currents (IL), converter currents (IC), source currents (IS)
Fig. 17. Harmonic spectrum of source current after
and DC-link voltage (VDC) under unbalanced linear load
compensation with ANFIS controller.
condition.

11
 Journal of Electrical Engineering
www.jee.ro

changed to two-phase load the source currents (IS) are still

maintained sinusoidal and balanced. It is noteworthy that,
both the controllers have good performance in load
balancing but the ANFIS controller has better DC voltage
regulation than the PI controller under unbalanced linear
load condition.

Fig. 18. Voltage regulation and load compensation with

PI controller under unbalanced load with voltage sag and
swell conditions.

A 65% three-phase voltage sag is applied in the

preferred feeder at t=0.05 s and ends at t=0.15 s. The
proposed SSTS control strategy detects the voltage sag,
which turned off the preferred feeder thyristor and turns Fig. 19. Voltage regulation and load compensation with
on the alternate feeder thyristors. It is observed from Figs. ANFIS controller under unbalanced load with voltage sag
18 and 19, during voltage sag, the alternate feeder gives and swell conditions.
full supply to the load. After the completion of voltage
sag duration, the alternate feeder supply gets turned off Neutral current will result in any circuit when there is
and the preferred feeder gives supply to the load. A 50% unbalanced flow of current i.e when the load in the three-
three-phase voltage swell is applied in the preferred phases are not balanced there is a net flow of current
feeder at t=0.35 s and ends at t=0.4 s. The proposed SSTS through the neutral line. Here the system is provided with
control strategy detects the voltage swell, which turned the unbalanced load and this leads to the flow of current
off the preferred feeder thyristor and turns ON the through the neutral. The suitable control strategy is
alternate feeder thyristors. It is observed from Figs. 18 adopted to cancel the neutral current. In order to achieve
and 19, during voltage swell, the alternate feeder gives zero neutral current at source side (ISn), the converter
full supply to the load. After the completion of voltage neutral current (ICn) is injected in opposite polarity with
swell duration, the alternate feeder supply gets turned off the load neutral current (ILn). It is observed from Fig. 20
and the preferred feeder gives supply to the load. the neutral current is fully eliminated and it is almost near
to zero for both PI and ANFIS controller.
Fig. 18 and Fig. 19 shows the performance of the PI
and ANFIS based DSTATCOM for load balancing along
with DC voltage regulation under unbalanced linear load
with voltage sag and swell conditions. It is observed from
the waveform of Figs. 18 and 19, the DSTATCOM
injected the appropriate current for each phase to make
the source currents sinusoidal balanced and in-phase with
the source voltage. It is also observed that, from 0.2 s to
0.3 s, when the three-phase unbalanced linear load is

12
 Journal of Electrical Engineering
www.jee.ro

(a) PI controller harmonic current and unbalanced source current. The load
compensation along with voltage regulation is achieved
by the combination of DSTATCOM with SSTS. Both
ANFIS and PI controller based DSTATCOM
compensates current harmonics successfully.
Nevertheless the ANFIS based DC-link voltage regulation
gives better performance than PI controller.

Acknowledgment
(b) ANFIS controller
The authors gratefully acknowledge the management
Fig. 20. Source neutral current compensation. of P. R. Engineering College Thanjavur and Thiagarajar
College of Engineering, Madurai, India, for providing all
6.3 Comparison of proposed work with existing work. the facilities to do the research work.
The Table 5 compares the THD compensation Appendix
performance of various control algorithms such as
symmetrical component (SC) theory, instantaneous Source Voltage 3-phase, 4-wire, 415V,50 Hz
reactive power (IRP) theory, instantaneous active and Line impedance Rs = 0.01 , Ls = 2.1 mH
reactive current component (IARCC) theory and the DC-link voltage of 700 V DC
proposed method. It is observed from this table the DSTATCOM
proposed ANFIS method offers better THD compensation DC-link capacitance of 3000 F
than other methods. DSTATCOM
DC-link voltage PI Kp=4.96 and Ki=1.01
Table 5 Comparison of %THD of source current with controller
existing and proposed method. Ripple filter Rf=10 , Cf=5 F
Load: R1=30 , L1=70 mH, R2=60
Existing methods Proposed Linear unbalanced , L2=100 mH, R3=120 ,
Methods load L3=400 mH.
Is Ph
3-phase uncontrolled bridge
IRP- IRP- IRP SC IARCC SRF- SRF rectifier,
PI ANFIS [32] [32] [32] PI ANFIS
[15] [15]
Nonlinear load RDC = 50 , LDC = 1 mH.

a 3.06 3.08 2.83 3.98 3.79 3.32 2.96 References

b 4.44 4.28 4.72 4.80 4.73 3.38 3.32 [1] A. Ghosh, G. Ledwich, Power quality enhancement
THD c 4.08 3.92 4.65 4.29 4.60 3.27 3.12 using custom power devices, Kluwer 2002.
%
[2] B. Singh, P. Jayaprakash, D.P. Kothari, New control
approach for capacitor supported DSTATCOM in three-
7. Conclusion phase four wire distribution system under non-ideal
supply voltage conditions based on synchronous reference
The performance of three-phase four-wire frame theory, Int. J. Electr Power Energy Syst. 33 (2011)
DSTATCOM along with SSTS for the mitigation of 1109-1117.
power quality issues like voltage sag/swell, neutral
current compensation, source harmonic reduction and
load balancing are extensively simulated in MATLAB/ [3] A. Shokri, H. Shareef, A. Mohamed, M. Farhoodnea,
SIMULINK. In order to verify the performance of SSTS, H. Zayandehroodi, A novel single-phase phase space-
the proposed system is analyzed with both voltage sag based voltage mode controller for distributed static
and swell conditions. It is clearly noticed that, the compensator to improve voltage profile of distribution
presented results have been confirmed the effectiveness of systems, Energy Convers. Manage. 79 (2014) 449455.
SSTS under both voltage disturbances. The SRF method
has been presented for the extraction of reference current [4] Seyedreza Aali, Unified Power Quality Conditioner
signals. The PV supported DSTATCOM injects current in Based on Neural-Network Controller for Mitigation of
such a way, to cancel out the excess neutral current,

13
 Journal of Electrical Engineering
www.jee.ro

Voltage and Current Source Harmonics, Journal of quality enhancement in three-phase four-wire distribution
Electrical Engineering, pp 1-6. system, Int. J. Hydrogen Energy 39 (2014) 19051-19066.

[5] R. Yuanjie, C. Li, Q. Ding, An adaptive harmonic [16] J.S.R Jang, ANFIS adaptive network based fuzzy
detection and a novel current control strategy for unified inference system, IEEE Trans. Syst. Man Cybern. 23
power quality conditioner, Simul. Modell. Pract. 17 (1993) 665-684.
(2009) 955-966.

[6] B. Singh, P. Jayaprakash, D.P. Kothari, A. Chandra, [17] S. Lalouni, D. Rekioua, T. Rekiou, E. Matagne,
K.A. Haddad, Comprehensive study of DSTATCOM Fuzzy logic control of stand-alone photovoltaic system
configurations, IEEE Trans. Ind. Inf. 10 (2014) 854-870. with battery storage, J. Power Sources 193 (2009) 899
907.
[7] D. Sreenivasarao, P. Agarwal, B. Das, Neutral current
compensation in three-phase, four-wire systems: A [18] N. Chettibi, Adel Mellit, FPGA-based real time
review, Electr. Power. Syst. Res. 86(2012) 170-180. simulation and control of grid-connected photovoltaic
systems, Simul. Modell. Pract. 43 (2014) 34-53.
[8] Chebabhi, Ali, et al. "Three Level Four Leg Shunt
Active Power Filter Based a New Three Dimensional [19] V. Kamatchi Kannan, N. Rengarajan, Photovoltaic
Space Vector Modulation strategy in the 0-axes." based distribution static compensator for power quality
Journal of Electrical Engineering, In Press, accepted improvement, Int. J. Electr Power Energy Syst. 42 (2012)
10.08 (2015). 685-692.

[9] H. Akagi, E. Hirokazu Watanabe, M. Aredes, [20] B. Singh, P. Jayaprakash, D.P. Kothari, A T-
Instantaneous power theory and applications to power Connected Transformer and Three-leg VSC Based
conditioning, Wiley-IEEE Press, 2007. DSTATCOM for Power Quality Improvement, IEEE
Trans. Power Electron. 23 (2008) 2710-2718.
[10] G. Benysek, M. Pasko, Power Theories for Improved
Power Quality, Springer, 2012. [21] P. Kumar, N. Kumar, A.K Akella, A simulation
based case study for control of DSTATCOM, ISA Trans.
53 (2014) 767775.
[11] V. Kamatchi Kannan, N. Rengarajan, Investigating
the performance of photovoltaic based DSTATCOM
using I cos algorithm, Int. J. Electr. Power Energy Syst. [22] P. Jayaprakash, B. Singh, D.P. Kothari, Digital
54 (2014) 376386. Signal Processor Implementation of Isolated Reduced-
rating Voltage Source Converter Using a Zig-zag
Transformer for Three-phase Four-wire Distribution
[12] R. Gupta, A. Ghosh, Frequency-Domain
Static Compensator, Electr Power Compon Syst. 39
Characterization of Sliding Mode Control of an Inverter
(2011) 15-30.
Used in DSTATCOM Application, IEEE Trans. Circuits
Syst. 53(2006) 662-676.
[23] Ishaque, Kashif, Zainal Salam, Hamed Taheri,
Modeling and simulation of photovoltaic (PV) system
[13] W.R.A Ibrahim, M.M Morcos, Artificial Intelligence
during partial shading based on a two-diode model,
and Advanced Mathematical Tools for Power Quality
Simul. Modell. Pract. 19 (2011) 1613-1626.
Applications: A Survey, IEEE Trans. Power Delivery 17
(2002) 668-673.
[24] N. Pandiarajan, R. Muthu, Mathematical Modeling
of Photovoltaic Module with Simulink, in: Proc of
[14] P. Kumar, A. Mahajan, Soft computing techniques
International Conference on Electrical Energy System,
for the control of an active power filter, IEEE Trans.
2011.
Power Delivery 24 (2009) 452-461.
[25] A. Mki, V. Valkealahti, Differentiation of multiple
[15] C.K. Sundarabalan, K. Selvi, PEM fuel cell
maximum power points of partially shaded photovoltaic
supported distribution static compensator for power
power generators, Renewable Energy 71 (2014) 89-99.

14
 Journal of Electrical Engineering
www.jee.ro

[26] W. Jiang, Y.F. Zhou, J.N. Chen, Modeling and

simulation of boost converter in CCM and DCM, in: Proc
of IEEE Conference on Power Electronics and Intelligent
Transportation System (PEITS), 2009.

[27] H. Mokhtari, S.B. Dewan, M.R. Iravani, Analysis of

a Static Transfer Switch With Respect to Transfer Time,
IEEE Trans. Power Delivery 17 (2002) 190-199.

[28] M.N. Moschakis, N.D. Hatziargyriou, A detailed

model for a thyristor-based static transfer switch, IEEE
Trans. Power Delivery 18 (2003) 1442-1449.

[29] S. Jazebi, S.H. Hosseinian, B. Vahidi, DSTATCOM

allocation in distribution networks considering
reconfiguration using differential evolution algorithm,
Energy Convers. Manage. 52 (2011) 27772783.

[30] B. Singh, S.K. Dube, S.R. Arya, An improved

control algorithm of DSTATCOM for power quality
improvement, Int J Electr Power Energy Syst. 64 (2015)
493504.

[31] W.J Xiao, Z.J. Xin, C.Y. Guang, T.Y. Heng,

Nonlinear modeling of a SOFC stack based on ANFIS
identification, Simul. Modell. Pract. 16 (2008) 399-409.

[32] T. Zaveri, B. Bhalja, N. Zaveri, Comparison of

control strategies for DSTATCOM in three-phase, four-
wire distribution system for power quality improvement
under various source voltage and load conditions, Int. J.
Electr Power Energy Syst. 43 (2012) 582-594.

15