[Downloaded from www.aece.ro on Tuesday, October 11, 2016 at 01:52:36 (UTC) by 39.32.114.62. Redistribution subject to AECE license or copyright.

Advances in Electrical and Computer Engineering Volume 12, Number 1, 2012

Phase Angle Control of Three Level Inverter

Based D-STATCOM Using Neuro-Fuzzy
Controller
Resul COTELI1, Erkan DENIZ1, Besir DANDIL1, Servet TUNCER2, Fikret ATA3
1
University of Firat, Technical Education Faculty, Electrical Science, Elazig, 23119, Turkey
2
Technical Education Faculty, Electronic and Computer Science,Elazig, 23119,Turkey
3
University of Firat, Electrical and Electronics Engineering,Elazig, 23119,Turkey
rcoteli@firat.edu.tr

1
Abstract—Distribution Static Compensator (D-STATCOM) phase angle control, direct current control, indirect current
is a shunt compensation device used to improve electric power control and constant dc-link voltage scheme are proposed
quality in distribution systems. It is well-known that D- for dynamic control of D-STATCOM’s output reactive
STATCOM is a nonlinear, semi-defined and time-varying
power [5, 6]. Phase angle control method has advantages
system. Therefore, control of D-STATCOM by the
conventional control techniques is very difficult task. In this such as its simple algorithm and easy to perform [2].
paper, the control of D-STATCOM is carried out by the neuro- Conventionally, phase angle control method is carried out
fuzzy controller (NFC) which has non-linear and robust by Proportional Integral (PI) controller designed according
structure. For this aim, an experimental setup based on three- to linear control techniques. In this control method, position
level H-bridge inverter is constructed. Phase angle control of characteristic roots on the complex plane is depended on
method is used for control of D-STATCOM’s output reactive
phase angle. To design a controller based on linear control
power. Control algorithm for this experimental setup is
prepared in MATLAB/Simulink and downloaded to DS1103 methods for phase angle controlled D-STATCOM,
controller card. A Mamdani type NFC is designed for control equations are linearized for specific operating point by some
of D-STATCOM’s reactive current. Output of NFC is assumptions. Nevertheless, it is impossible to free the
integrated to increase tracking performance of controller in position of characteristic roots from phase angle in real-time
steady state. The performance of D-STATCOM is applications. In addition, such controllers may not give the
experimentally evaluated by changing reference reactive
desired performance due to parameter variation, load
current as on-line. The experimental results show that the
proposed controller gives very satisfactory performance under disturbance, unmodelled dynamics of the plant [7, 8].
different loading conditions. To cope with disadvantages of linear control methods,
some researchers suggested non-linear control methods such
Index Terms—electric power quality, D-STATCOM, three- as H ∞ , differential algebra theory, exact linearization with
level H-bridge inverter, neuro-fuzzy controller, MATLAB. feedback [8-12]. Such control methods need exact
mathematical model and parameters of controlled system.
I. INTRODUCTION Besides, controller design with these methods has a more
Recently, electrical power quality problems have been complex structure and is more difficult to realize than linear
increased as a result of widespread use of power controllers [8].
semiconductor components. There are many methods for the To obtain a good performance both in transient and steady
solution of the power quality problems in distribution state from D-STATCOM, controller should be nonlinear and
systems such as using surge arresters, active power filters, robust structure and independent from mathematical model
isolation transformers, uninterruptible power supplies and of controlled system. Recently, intelligent controllers as
static VAR compensators [1]. The inverter based custom alternative linear and nonlinear control techniques have been
power devices have been widely used in many utilities used in the control of D-STATCOM [13-18]. In case of
because of their fast response and good steady state using the intelligent controllers, there is no need the
performance over conventional compensators [2]. The D- mathematical model of the controlled system. Moreover,
STATCOM is an inverter based custom power device used such controllers can provide efficient control over wide
for the voltage regulation, power factor correction and load operating conditions which distinguishes them from
balancing. conventional linear controllers [8]. NFC is an intelligent
Control method has a vital important over the whole controller that is based on Fuzzy Logic Controller (FLC)
performance of the D-STATCOM [3]. The system are whose functions are realized by Artificial Neural Network
controlled to obtain desired performance depending on the (ANN) and it has been successfully applied to the most of
operational requirements, type of applications, system control problems. NFCs are substantially effective in control
configuration and loss optimisation [3]. The control method of nonlinear systems [19]. In [20], NFC is used for current
to be preferred for D-STATCOM is simple its algorithm and control of D-STATCOM using indirect current control
easy to implement [4]. Different control methods such as method. But, experimental results with this kind of control
method haven’t been reported previously for D-STATCOM
1
This work was supported by Scientific and Technology Research in literature.
Council of Turkey (TUBITAK) with 107E245 numbered project.
Digital Object Identifier 10.4316/AECE.2012.01013

77
1582-7445 © 2012 AECE
 [Downloaded from www.aece.ro on Tuesday, October 11, 2016 at 01:52:36 (UTC) by 39.32.114.62. Redistribution subject to AECE license or copyright.]

Advances in Electrical and Computer Engineering Volume 12, Number 1, 2012

In this paper, the robust nonlinear control strategy based III. THREE-LEVEL INVERTER BASED D-STATCOM
NFC with fuzzy singleton rules is proposed for phase angle Inverters are convetionally controlled as a PWM voltage
control of D-STATCOM to achieve a good dynamic source in medium and high-power applications because of
response under the all possible loading conditions. Designed switching losess. Since the converter is operated as a PWM
NFC has two inputs, single output and four layers. Inputs of voltage source, the switching frequency of the devices can
NFC are chosen as error of q-axis current and variation of be properly controlled. However, harmonic distortion in the
this error. An anti wind-up integrator is used to improve generated reactive current is much more than that of a
tracking capability of NFC in steady state. The gradient controlled current scheme. To reduce low-order harmonics,
descent algorithm is used to train the NFC in direct adaptive either inverter have been operated in parallel or multilevel
control scheme using the simulation model of the D- converters should be used. In a inverter structure operated in
STATCOM and then trained NFC is used in experimental parallel, the input transformer becomes bulky [21].
study. This paper is organized as follows. In Section 2, Recently, multilevel inverters have been widely used in fast
operation principle of D-STATCOM is summarized. Power controllable reactive power applications. These inverter
circuit and mathematical model of three-level H-bridge topologies used in reactive power applications offer
inverter based D-STATCOM is given in Section 3. The considerable advantages over parallel converter operations
description of the NFC and its training algorithm are in terms of space reductions, cost and losess. Cascaded
explained in Section 4. The experimental results related to inverter is the most popular topology among multilevel
proposed controller are given in Section 5. Section 6 inverters because of using minimum number of components
provides conclusions of this study and its modular structure [22]. The number of lowest
voltage-level in the cascaded inverters is three. In addition,
II. BASIC OPERATION PRINCIPLE OF D-STATCOM dc voltage unbalance doesn’t occurred in three-level. In this
The basic operation principle of D-STATCOM can be study, three-level H-bridge inverter is preffered for power
explained by Fig. 1. From this figure, D-STATCOM is circuit of D-STATCOM. Flexible structure of this inverter
connected to Bus2. The amount of the reactive power easily allows to increase the number of voltage-level by
flowing from Bus1 to Bus2 mainly depends on the voltage adding the H-bridges in series to per phase.
difference between both buses. Reactive power flows from Power circuit of three-level H-bridge inverter based D-
the higher voltage to low voltage. In the phase angle control STATCOM is illustrated in Fig. 2. D-STATCOM consists
method, the amplitude of inverter output voltage is of three H-bridge inverters, three dc-link capacitors (C)
controlled by causing a small amount of active power flow supplying the dc voltages to H-bridge inverters and a
into/out of the D-STATCOM. If V i lags V s , a small amount coupling inductance with internal resistance (L s + R s )
of active power flows into D-STATCOM and dc-link connecting to ac grid the inverter.
capacitor voltage increases by charged. On the contrary, a
small amount of active power flows out D-STATCOM, thus
the dc-link capacitor voltage decrease by discharged.
Briefly, dc-link voltage is controlled by adjusting phase
angle  between the V s and V i voltages.

Figure 1. Simple two-bus system in which D-STATCOM is connected

Figure 2. Power circuit of three-level H-bridge inverter based D-STATCOM

The amount of active power flow between D-STATCOM Mathematical model of H-bridge inverter based D-
and ac grid is expressed as Eq.(1). STATCOM can be derived from Fig. 1. The circuit equation
on stationary reference frame can be written as follows,
V s Vi di
P sin  (1) Ls  Rs i  V s  V i (3)
X dt
The large amount of reactive power generated/absorbed
The V s is defined as complex voltage vector of grid and
by the D-STATCOM can be also controlled by flowing the
small amount of active power. The amount of reactive the V i is also defined as complex voltage vector of inverter.
power generated/absorbed by D-STATCOM is also given in Space vector theory based on synchronous reference frame
(2); is applied to D-STATCOM. Coordination of vector space is
shown in Fig. 3. The and dq-axes represent stationary
Vs reference and synchronous rotating frame, respectively.
Q (Vs  Vi cos ) (2)
X Voltage vectors and current vector in complex plane are
expressed in stationary reference frame as follows [23];

78
 [Downloaded from www.aece.ro on Tuesday, October 11, 2016 at 01:52:36 (UTC) by 39.32.114.62. Redistribution subject to AECE license or copyright.]

Advances in Electrical and Computer Engineering Volume 12, Number 1, 2012

V s  Vsa  V sb e j  V sc e 2 j  Vs e j (4) By substituting (7), (8), (9) into (10), the circuit equations
in dq-axis are obtained as (11) and (12).
V i  Via  Vib e j  Vic e 2 j  Vi e j (  ) (5) di d
Ls  Rs id  v sd  vi cos   Ls iq (11)
dt
i  i a  ib e j  ic e 2 j  ie j (   ) (6) diq
Ls  Rs iq  v sq  vi sin   Ls id (12)
Where,   2 / 3 , V sabc and V iabc are instantaneous phase dt
voltages of grid and D-STATCOM, respectively. The d-axis Where,  is frequency of grid. Peak value of inverter
of the space vector diagram shown in Fig. 3 is assigned to output voltage is expressed as (13).
coincide with the V i . Complex vectors in the stationary
reference frame are transformed onto synchronous rotating V  M aVdc (13)
frame by multiplying them with unity space vector e  j as In (13), M a is modulation index and V dc is dc-link
follows [23]; voltage. From instantaneous power equality on the dc side
 and the ac side of the inverter, power equation can be
written as (by neglecting the system losses) [24];
 3
Pc  V dc I dc  (vid I d  viq I q ) (14)
2
Where I dc is the capacitor current. Capacitor current is

 also descibed as follows,

3 dV
 I dc  M a (i d cos   i q sin )  C dc (15)
2 dt
Figure 3. Coordinate system for synchronous rotating reference frame State equations for ac side of three-level H-bridge inverter
based D-STATCOM are given by (11) and (12) whereas
di
L s e  j  R s e  j i  V s e  j  V i e  j (7) Eq.(15) presents state equation for dc side of three-level H-
dt bridge inverter based D-STATCOM. ac and dc side circuit
equations in dq-axis can be rearranged in state space matrix
V s e  j  v sd  jv sq (8)
form as (16). It is shown that circuit equations of D-
STATCOM are nonlinear if  is taken as input variable. The
V i e  j  v id  jv iq  vi cos   jv i sin  (9)
circuit equations should be linearized around the certain
operation point for controller design based on linear control
i e  j  i d  j i q (10) method. Assuming that V sd and V sq are constant value,
linearization of Eq.(16) around the operation point  
presents a set of linear equations given in (17) [23].
 
 i d    R s / Ls   M a cos  / Ls   i d  1 / Ls 0 
d    v sd 
iq    Rs / Ls M a sin  / Ls    i q    0 1 / Ls     (16)
dt    3M cos   v sq
Vdc   a

3M a sin 
0  Vdc   0 0   
 2C 2C 

 M 
   ( a V dc 0 sin  0 ) 
 id    R s / Ls   M a cos  0 / L s   i d   Ls

d    M
iq    R s / Ls M a sin  0 / L s    i q    ( a V dc 0 cos  0 )  (17)
dt    3M cos  3M a sin  0
  Ls 
V dc   a 0
 0  V dc   3 
   M
2C 2C a
(i d 0 sin  0  i q 0 cos  0 ) 
 2C 

The characteristic equation of system given by (17) is

IV. STRUCTURE OF NEURO-FUZZY CONTROLLER
expressed by (18).
FLC is able to make inferences from an expert
 R  2 3M 2  3M 2 R knowledge. ANN has features such as learning,
R  
s 3  2 s s 2   s   a
  2 s  a
2
s
(18) generalization and adaptation. NFC is an intelligent control
Ls L
 s  2 L s C  2 L s C
method combining the features of FLC and ANN. Basically,
NFC is based on the principle which the functions of FLC
are implemented by ANN [25]. Therefore, NFC shows
features of a nonlinear controller that has abilities such as

79
 [Downloaded from www.aece.ro on Tuesday, October 11, 2016 at 01:52:36 (UTC) by 39.32.114.62. Redistribution subject to AECE license or copyright.]

Advances in Electrical and Computer Engineering Volume 12, Number 1, 2012

learning, adaptation and making inferences. In addition, the membership value for the change in current error. For
NFC doesn’t need mathematical model of the system to be the kth rule node;
controlled. Mentioned features make NFC superior when
compared with other controllers. net k3   w3jk x 3j , y k3  net k3 (23)
j
In this study, parameter learning of NFC with fuzzy logic
rules whose consequents are fuzzy singleton and its network Where, x 3j represents the jth input to the node of rule
architechture are shown in Fig. 4. Fuzzy singleton rules are
layer, and w3jk is assumed to be unity. The output layer acts
in the following form.
a defuzifier. The single node in this layer collects all
Rj: IF x 1 is A1j AND x 2 is A 2j THEN y is w j (19) incoming signals from the rule layer to obtain the final
results.
j
Where, x i is an input variable, y is output variable, A i is
linguistic terms of precondition part with membership net o4   wko
4 3
yk (24)
k
function  j ( x i ) , w j is real number of the consequent part,
Ai 4
Where, the links weights wko represent the output action
j=1, 2, …M and i=1, 2, ..n.
of the kth rule. The output of the system using the central
w 3jk defuzzification for the Mamdani fuzzy model is given as;
y 2j y 3k
w 2
ij
net o4
y 1i w 4 y o4  (25)
x1  e ko  yk3
k

Steady state error in the output of NFC is eliminated by

y o4
using external compensator [20]. Thus, the desired phase
angle is obtained from integrated outputs of controller.
The NFC can adjust the fuzzy control rules by modifying
x2  Δe
the output weights. The parameters of the membership
functions and the output weights of NFC are modified using
the back-propagation algorithm to minimize the
performance index E [26];

E  0.5.e 2 (26)

Figure 4. Architecture of NFC The parameters w ko in Eq.(24) can be modified as follows

[27].
As shown in Fig.4, the NFC consists of four layers. In
E
Layer-I, the input variables of NFC are presented and they wko  wko   (27)
are given by (20) and (21). wko

e(k )  iqref (k )  iq (k ) (20) Where, the η is the learning rate. The gradient of the
performance index can be derived as follows.
e(k )  e(k )  e(k  1) (21)
E iq y o4 neto4 y k3
Where, e, i qref - i q , is the current tracking error and e is  e. sgn( 4 )  δ (28)
wko y o neto4 wko  yk
3
variation of current error. , i qref and i q are also the reference k
and actual currents for q-axis, respectively.
The Layer-II calculates the degree of membership Where,  i q / y o4 can be calculated using the dynamics of
functions for the input values. The Gaussian activation D-STATCOM. Gradient of the performance index for the
function are utilized to represent the membership functions. parameters of the membership functions can be also derived
The weights between the input and membership layers are as (29) and (30).
assumed to unity. The output of this layer is expressed as;
E Δi q 1 4
xi  mij 2
(xi  mij ) 2  e.sgn ( ) wko yj (29)
net 2j   , y 2j  exp (net 2j ) (22) mij 4 3
Δy o  y k (σ ij ) 2
2(σ ij ) 2 k

Where,  ij and m ij are the standard deviation and mean E Δiq 1 4

(xi  mij ) 2 2
 e.sgn ( ) wko y j (30)
σ ij Δy o4  y k3 (σ ij )3
of the Gaussian function in the jth term of the ith input node. k
The Layer-III of the NFC includes the fuzzy rule base
and the nodes in this layer represented by determine the IV. CONTROL STRATEGY OF THREE-LEVEL INVERTER BASED
fuzzy rules. Each node takes two inputs, one from the D-STATCOM
membership value for the current error, and the other from
In phase angle control method, modulation index is kept
constant and the fundamental voltage component of inverter

80
 [Downloaded from www.aece.ro on Tuesday, October 11, 2016 at 01:52:36 (UTC) by 39.32.114.62. Redistribution subject to AECE license or copyright.]

Advances in Electrical and Computer Engineering Volume 12, Number 1, 2012

is controlled by dc-link voltage. The control variable is obtained. Amplitude of modulation signals M a is kept
therefore δ. In order to change the amount of reactive power constant. Thus, modulation signals for phase-abc are
delivered/absorbed by D-STATCOM, dc-link voltage obtained as M a sin(+), M a sin(+ and
should be adjusted by control of δ [28]. Fig. 5 shows phase M a sin(+), respectively. Then, PWM signals are
angle control diagram of D-STATCOM in which phase produced by three level Sinusoidal Pulse Width Modulation
angle is controlled by NFC. A Software Phase Locked Loop (SPWM) technique. In three-level SPWM, two carriers with
(SPLL) is used to synchronize inverter output voltage with same frequency and amplitude are used. The reference
ac grid. Sensed currents and voltages in point of common waveform is centered in the middle of the carrier set and it is
coupling (PCC) are transformed to dq- axis compenents by continuously compared with each of the carrier signals. If
using grid frequency obtained from SPLL. The reactive the reference is greater than carrier signal, then the active
current error and variation of this error are applied to NFC. device corresponding to that carrier is switched on.
The integrated output of NFC is reference signal Otherwise, the active device corresponding to that carrier is
representing . Thus, phase angle of modulation signals are switched off.
Three-Level H-
Bridge Inverter

Va H-bridge
C
inverter-1
AC Grid PCC Rs Ls
Vb H-bridge
C
inverter-2
iabc Vabc SPLL
Vc H-bridge
C
inverter-3
abc/dq

PWM signals
Vsdq-Isdq NFC
Reactive power
calculation K Ts
de/dt z-1
ML-SPWM
Qref

Figure 5. Phase angle control diagram with NFC for D-STATCOM

V. EXPERIMENTAL RESULTS to ac grid via a coupling inductance with internal resistance.

A DSP-based experimental setup of 8.5kVA is built to A blanking circuit is used to prevent short circuit of
validate the operation of the D-STATCOM using NFC. The switches on the same leg. The current and voltage signals of
laboratory setup is shown in Fig. 6. The system can be point of common coupling are sensed by using Hall effect
divided into four parts as power circuit, controller card, LEM tranducers (LA25-SP1 and LV1000), respectively.
measurement&protection circuits and gate drive circuits. A During the real-time operation, the sensed current and
personal computer is used to program the required control voltage signals are processed by using the proposed control
algoritm and load the control algorithm to the DS1103 algorithm and produced PWM signals are applied to the gate
controller card. Three-level H-bridge inverter is constructed drive board. The circuit parameters related to operating
by PM75CLA120 IPM modules. This inverter is connected condition of D-STATCOM are given in Table I.

Figure 6. Experimental setup of D-STATCOM

81
 [Downloaded from www.aece.ro on Tuesday, October 11, 2016 at 01:52:36 (UTC) by 39.32.114.62. Redistribution subject to AECE license or copyright.]

Advances in Electrical and Computer Engineering Volume 12, Number 1, 2012

TABLE I. CIRCUIT PARAMETERS RELATED TO OPERATING

CONDITION OF EXPERIMENTAL SETUP

Phase voltage of grid : 200V

Coupling reactance : 0.1 Ω +j2.89mH
dc-link capacitors : 3.3mF/450V
Switching frequency : 1.25kHz
Dead time for each switch : 4.5µs
Sampling time : 100µs

A three-phase variac is inserted between the coupling

inductance and the ac grid. Before start-up, the output Figure 7. The variations in voltages of three dc-link capacitors when D-
voltage of the variac is adjusted to phase voltage of 200V to STATCOM was put into operation
charge the dc capacitors. In addition, a resistor is connected
in series with the dc capacitors to limit the charging current Figure 8(a) and (b) show the reactive current response of
and then they are bypassed after initial charging. Inverter is D-STATCOM to step changes in the reactive current
controlled by ML-SPWM using fixed switching frequency command from positive to negative and for opposite
of 1.25kHz. Laboratory tests of the prototype D-STATCOM direction, respectively. To provide clear view, Fig. 8(a1) and
are performed to verify the validity of proposed NFC (b1) are also zoomed as shown in Fig. 8(a2) and (b2). It can
controller. Fig. 7 shows the experimental waveform related be seen that D-STATCOM generates/absorbs the reference
to voltages of three dc-link capacitors when the D- reactive current in a small period of time for two reference
STATCOM is started in stand-by mode. From the figure, changing. In addition, there is no steady state error in
three dc-link capacitors are firstly charged to about 245V in reactive current.
phase voltage of 200V. Voltages of three dc-link capacitors Figure 9(a) and (b) show the changing of three dc-link
increase up to 310V after gating signals are applied to voltages for same command changes in the reactive current
switches of inverter. In addition, three dc-link voltage are command. It is clearly seen that three dc-link voltages have
balanced. small overshoots and reach their reference value quickly. In
The task of the D-STATCOM is mainly to provide the addition, dc-link voltages increase up 350V for capacitive
reactive current for load compensation and voltage operation mode of D-STATCOM whereas dc-link voltages
regulation and suppress current harmonics. In this study, D- decrease about 270V for inductive operation mode of D-
STATCOM is operated in reactive current source mode. For STATCOM.
testing of D-STATCOM, a step command for reference The phase-a voltage and current of D-STATCOM are
reactive current is used to emulate sudden load or voltage given in Fig. 10(a) and (b) as per-unit (pu). Phase-a current
changing in system. This case is the worst operating of D-STATCOM is leading about 90° respect to its voltage
condition for D-STATCOM. During test, reactive current for capacitive mode. Otherwise, phase-a current of D-
and voltages of three dc-link capacitors are observed. For STATCOM is laging about 90° respect to its voltage for
reactive current and three dc-link capacitors voltages, inductive mode.
performance of D-STATCOM is evaluated for settling time,
reference tracking and steady state error.

(a1) (a2)

82
 [Downloaded from www.aece.ro on Tuesday, October 11, 2016 at 01:52:36 (UTC) by 39.32.114.62. Redistribution subject to AECE license or copyright.]

Advances in Electrical and Computer Engineering Volume 12, Number 1, 2012

(b1) (b2)

Figure 8. Reactive current response of D-STATCOM, (a) for changing of i qref 0A to +20A, (b) for changing of i qref 0A to -20A

(a) (b)

Figure 9. The variations in voltages of three dc-link capacitors for changing of reference reactive current, (a) 0A to+20A, (b) 0A to -20A

(a) (b)
Figure 10. Relation between phase-a current and voltage for a step change of reference reactive current, (a) 0A to +20A, (b) 0A to -20A

83
 [Downloaded from www.aece.ro on Tuesday, October 11, 2016 at 01:52:36 (UTC) by 39.32.114.62. Redistribution subject to AECE license or copyright.]

Advances in Electrical and Computer Engineering Volume 12, Number 1, 2012

[12] X. Yang, H. Hao and Y. Zhong, “The State-Space Modeling and

VI. CONCLUSION Nonlinear Control Strategies of Multilevel DSTATCOM”, Power and
Energy Engineering Conference (APPEEC), 2010, pp. 1 – 4.
In this paper, the application of phase angle controlled D- [13] M. P. Kazmierkowski and L. Malesani, “Current Control Technique
STATCOM using neural fuzzy networks with singleton for Three-Phase Voltage Source PWM Converters”, IEEE Trans. Ind.
rules is presented. The NFC is preffered due to its non-linear Electron., vol. 45, no. 5, pp. 691–703, 1998.
[14] L. Malesani and P. Tomasin, “PWM Current Control Techniques of
and robust structure. In addition, output of NFC is integrated Voltage Source Converters–A Survey”, Int. Conf. Proc. Industrial
to improve tracking capability in steady state error. Thus, a Electronics, Control and Instrumentation, IECON’93, 1993, pp. 670–
good control performance is obtained from D-STATCOM 675.
both in transient and steady state conditions. The validaty of [15] C. Li, Q. Jiang, X. Xie and Z. Wang, “Rule-Based Control for
STATCOM to Increase Power System Stability”, Proc.Int. Conf.
the proposed control system is shown experimentally for Power System Technology, POWERCON, 1998, pp. 372–376.
changes of reactive power demand. From the experimental [16] C. Li, Q. Jiang, Z. Wang and,D. Retzmann, “Design of A Rule-Based
results, very quick response time and accurate compensation Controller for STATCOM”, Proc. Industrial Electronics Society,
1998, pp. 467–472.
effect are obtained by using the proposed control method [17] W. Chen, Y. Liu, J. Chen and J. Wu, “Control of Advanced Static
with NFC. VAR Generator by Using Recurrent Neural Networks”, Proc. Power
System Technology, POWERCON’98, 1998, pp. 839–842.
[18] M. Mohaddes, A. M. Gole and P. G. Mclaren, “A Neural Network
VII. REFERENCES Controlled Optimal Pulse-Width Modulated STATCOM”, IEEE
[1] H. Masdi, N. Mariun, S. M. Bashi, A. Mohamed and S. Yusuf, Trans. Power Deliv., vol. 14, no. 2, pp. 481–488, 1999.
“Construction of a Prototype D-STATCOM for Voltage Sag [19] B. Dandil, “Fuzzy Neural Network IP Controller For Robust Position
Mitigation”, European Journal of Scientific Research’, vol. 30, no. 1, Control of Induction Motor Drive”, Expert Systems with Applications,
pp.112-127, 2009. vol. 36, pp. 528-4534, 2009.
[2] D. Masand, S. Jain and G. Agnihotri, “Control Algorithms for [20] E. Deniz, R. Coteli, B. Dandil, S. Tuncer, F. Ata and M. T. Gencoglu,
Distribution Static Compensator”, Industrial Electronics IEEE “Neuro-Fuzzy Current Controller for Three-Level Cascade Inverter
International Symposium, 2006, pp. 1830-1834. Based D-STATCOM", UPEC'2010: 45th International Universities’
[3] B. Singh, R. Saha, A. Chandra, and K. Al-Haddad, “Static Power Engineering Conference, Cardiff, 31st August- 3rd September
Synchronous Compensators (STATCOM): A Review”, IET Power 2010, pp. 1-5.
Electronics, vol. 2, pp. 297-324, 2009. [21] K. Chatterjee, D. V. Ghodke, A. Chandra and K. Al-Haddad, “Simple
[4] B. Singh and J. Solanki, “An Improved Control Approach for Controller for STATCOM-Based Var Generators”, IET Power
DSTATCOM with Distorted and Unbalanced AC Mains”, Journal of Electronics, vol: 2, pp. 192-202, 2009.
Power Electronics, vol. 8, no. 2, pp.131-140, 2008. [22] S. Sirisukprasert, “The Modelling and Control of a Cascaded-
[5] A. Cetin, “Design and Implementation of a Voltage Source Converter Multilevel Converter-Based STATCOM”, PhD. Thesis, Dept. ECE
Based STATCOM for Reactive Power Compensation and Harmonic Virginia Tech., USA, 2004.
Filtering”, PhD. Thesis, The Graduate School of Natural and Applied [23] N. Voraphonpiout, T. Bunyagul and S. Chatratana, “Analysis and
Sciences of METU, Ankara, 2007. Performance Investigation of a Cascaded Multilevel STATCOM for
[6] S. Kumar, “Static Synchronous Compensators at Distribution and Power System Voltage Regulation”, International Energy
Transmission Levels”, Lecture Note, 2003. Journal, vol. 8, no. 2, 2007.
[7] S. H. Hosseini, R. Rahnavard and Y. Ebrahimi, “Reactive Power [24] P. Giroux, G. Sybille and H. Le-Huy, “Modeling and Simulation of a
Compensation in Distribution Networks with STATCOM by Fuzzy Distribution STATCOM using Simulink’s Power System Blockset”,
Logic Theory Application”, Power Electronics and Motion Control IECON’01: The 27th Annual Conference of the IEEE Industrial
Conference, 2006, pp.1-5. Electronics Society, 2001, pp. 990-994.
[8] S. Mohagheghi, “Adaptive Critic Designs Based Neuro-controllers for [25] J. S. R. Jang, C. T. Sun and E. Mizutani, “Neuro-Fuzzy and Soft
Local and Wide Area Control of a Multimachine Power System with Computing”, Prentice Hall, USA, 1997.
a Static Compensator”, PhD Thesis, Georgia Institute of Technology, [26] M. Gökbulut, B. Dandil and C. Bal, “Development and
2006. Implementation of Fuzzy-Neural Network Controller for Brushless
[9] F. Liu, S. Mei, Q. Lu, Y. Ni, F.F Wu and A. Yokoyama, A., “The DC Motors”, Intelligent Automation and Soft Computing, An
Nonlinear Internal Control of STATCOM: Theory and Application”, International Journal, vol. 13, no. 4, pp. 423-435, 2007.
International Journal of Electrical Power & Energy Systems, vol. 25, [27] M. Gökbulut, B. Dandil and C. Bal, “A Hybrid Neuro-Fuzzy
pp. 421 – 430, 2003. Controller for Brushless DC Drives”, Lecture Notes in Artificial
[10] P. Petitclair, S. Bacha and J. P. Rognon, “Averaged Modelling and Intelligence, vol: 3949, pp. 125-132, 2006.
Nonlinear Control of an Advanced Static Var Compensator”, [28] A. Cetin and M. Ermis, “VSC-Based D-STATCOM With Selective
Proceedings of the Power Electronics Specialists Conference, 1996, Harmonic Elimination”, IEEE Transactions on Industry Applications,
pp. 753-758. vol. 45 , pp. 1000 – 1015, 2009.
[11] Z. Yao, P. Kesimpar, V. Donescu, N. Uchevin and V. Rajagopalan,
“Nonlinear Control for STATCOM Based on Differential Algebra”,
29th Annual IEEE Power Electronics Conference, 1998, pp. 329-334.

84