THE IMPACT OF A FUZZY SELF-TUNING PI CONTROLLER

REINFORCED BY STATCOM ON LCC-HVDC SYSTEM CONNECTED TO

WEAK AC GRID IN BOTH SIDE

S. KHERFANE S.HADJERI
ICEPS Laboratory, Department of Electrical Engineering,
University Djillali Liabes of Sidi Bel Abbes, PB 89 Sidi Bel Abbes 22000, Algeria
sam1975mir@gmail.com, samir.hadjeri@univ-sba.dz

S.A. ZIDI
ICEPS Laboratory, Department of Electrical Engineering,
University Djillali Liabes of Sidi Bel Abbes, PB 89 Sidi Bel Abbes 22000, Algeria
sbzidi@yahoo.fr

Abstract: The AC/DC interaction between an HVDC Generally, the HVDC systems are used
system and the adjacent weak AC system is very sensitive frequently for long distance transmission, because of
to disturbances. For this reason, This paper presents an their electrical losses opposite to alternative current.
effective control system integrated in HVDC system, Sometimes, the HVDC for shorter distances is
based on the use of different values of proportional and
integrals gains varied online by Fuzzy Self-Tuning PI necessary despite the higher costs because of other
Controller (FSTPIC) according to necessity, supported by advantages obtained by this connection, such as
a static synchronous compensator (STATCOM) to improved system stability and interconnection
compensated the reactive power and achieve improved between asynchronous AC systems. [2]
performance of the HVDC system. The fuzzy logic device HVDC is actually part of the electric grid in
having self-adaptive gains of inputs to ensure their different places of the power transmission system
normalization between -1 and +1. and therefore interaction imposes itself. When a
To evaluate the effectiveness of the proposed device, a HVDC transmission system is connected to a weak
comparison is made between four system of CIGRE AC system, The AC/DC interaction becomes more
benchmark model, the first is conventional system (PI
controller only), the second HVDC system with Fuzzy
sensitive to disturbances and the search for solutions
Self-Tuning PI Controller, the third is HVDC system with to control it is inevitable. The general objective is
the STATCOM only and finally HVDC system with Fuzzy the construction of a valid and appropriate strategy
Self-Tuning PI controller and the STATCOM. This of control, for the different operating conditions, so
comparison is done to different types of default as a we must respect the equilibrium between speed
single-phase short circuit to ground and three-phase response and stability, if there are small disturbances
short circuit to ground inverter side of HVDC on the one hand, and to support the disturbances due
transmission link system. to faults and the switching on the other hand. In
addition, the highly non-linear nature of the control
Key words: LCC-HVDC system, Fuzzy Self-Tuning PI loops requires a wide parameter control which takes
Controller, STATCOM, CIGRE Benchmark model,
Commutation Failure. into account a range of operating conditions.[3]
Recently, intelligent control systems such as
1. Introduction fuzzy logic controllers are applied to HVDC systems
Today, the production, and distribution of the AC to dampen dynamic oscillations [4]. Actually, these
energy is selected for the following main reasons: controllers show good control performance when the
the facility of production, the voltage change and systems are complex and cannot be analyzed.
easiness of cutoff. However, control of transferring The nature of the converters of an HVDC link
energy AC power through networks poses problems based on the power electronics which to a non-linear
more difficult to solve. There is, situations in which behavior, leads to undesired Increment, from
the transmission of electrical energy into high current harmonics and reactive power in AC to weak
voltage direct current (HVDC) is more interesting power factor and overall weakness of performance.
than the alternative and more economical power, for The STATCOM (synchronous static compensator) is
example, if the reactive power compensation and developed as a reliable method to compensate for
stability become difficult to ensure. But this type of the problems mentioned. [5]
power transmission remains a concern power Therefore, it is clear that an HVDC system
transmission only and not the production or supported by a Fuzzy Self-Tuning PI controller and
distribution, it has advantages; depends on the STATCOM can be given good results in order to
command and control of very reliably.[1] find an adequate solution for that will test this
proposal device.
 Fig. 1. Schematic diagram of test system
2. The Test System The direct current Id flowing from the rectifier to
The test system in this paper is the first CIGRE the inverter is:
HVDC benchmark system supported by (FSTPIC)
and STATCOM illustrated in Fig. 1. The simulation (Vd 0 r Cos  Vd 0i Cos )
of this system is realized in Matlab/Simulink . Id  (1)
The HVDC system is mono-polar 500kV, 1000 Rcr  RL  Rci
MW HVDC link with a twelve pulse converters on
each sides (rectifier and inverter). Each converter Where Vd0r,Vd0i, RL, RCr and RCi are an open circuit
station has a 12-pulse thyristor based line- rectifier DC voltage, an open circuit inverter DC
commutated converter (LCC) which is formed by voltage, resistance of DC line, equivalent
two six pulse Graetz converter bridges. Both side of resistances of the rectifier switching and equivalent
HVDC are connected to the weak AC system, which resistances of inverter switching, respectively.
is a good choice for the test system in HVDC control
studies [6]. The short circuit ratios (SCRs) for the According to (1), a small variation of Vd0r or Vd0i
CIGRE model are: voltages can cause a very large change of the direct
current Id because the resistance of the DC line and
• Rectifier: 2.5∠84° at 50Hz. the other resistors of the converters are relatively
• Inverter: 2.5∠75° at 50 Hz low. It is essential to design a control system to
solve this problem of instability. [9-10]
The data system and a detailed model for the
HVDC system can be found in [7-8]. 4. Modes of Control of HVDC System
The action on the firing angles of the valves of
3. Basic Principles the converters able to regulate these voltages.
The essential role of control systems is to The control functions which enable the
distribute a sequence synchronous pulse to the adjustment of the converter will be ensured at the
valves AC / DC of converter station network that level of the converter according to the following
puts them in conduction. modes of control:
The sequence should be highly regularity to
reduce operating disequilibrium and eliminate A. control in rectifier
harmonics and to allow the voltage controllability 1) Control with minimal firing angle mode:
straightened. The power flows adjustment is based To avoid commutation failure must ensure a
on the voltage setting across the rectifier or inverter minimum value of firing angle min of the rectifier,
in an increase or decrease. This adjustment is made to ensure sufficient voltage across the valves before
with a quick adjustment on the firing angles of the these gates get the firing order.
valves. Fig. 2. shown the equivalent circuit of a 2) Control with constant curent (CC) mode:
single-pole HVDC link. Using a control loop that ensures the increase of
the firing angle  if the measured current (Idmes) is
greater than the value of the reference current (Idref)
Rcr RL -Rci
and the decrease if the contrary
Id
B. control in inverter:
Vd0r Cos  Vdr Vdi Vd0r Cos 
1) Control with Constant Extinction Angle
(CEA) mode:
The setting mode to the inverter is made to
determine its firing angle, which allows to obtain an

Fig. 2. The equivalent circuit of a single-pole HVDC link.

extinction angle  desired by a control loop that between reference greatness and measured
compares the angles prior extinctions (mes) to a greatness, previous value of error, the error rate, the
reference set (ref) and makes in the right direction. reference greatness, the measured greatness and the
sampling time, respectively.
5. Interaction between AC and DC system
The study of the interaction between high voltage gref PI U
System
power systems (HVDC) and alternative networks is Controller

necessary to improve the performance of HVDC gmeas

power system, as stable voltages, overvoltage, E

Fuzzy
Kp Ki

resonances and recovery of normal operation after du/dt

R Controller

disturbance.
The degree of AC / DC interaction depends on Fig. 3. Block diagram of Fuzzy-PI controller.
the capacity of the adjacent AC network of the
HVDC link to the transmitted DC power. The inputs (E and R) and the outputs (Kp and
Ki) from the fuzzy schemes are fuzzified into seven
A. The strength of a conversion system sets, namely, PB: Positif Big, PM: Positif Medium,
Therefore, you have to characterize the strength PS: Positif Small, Z: zero, NS: Negatif Small, NM:
of a conversion system AC / DC, to determine the Negatif Medium, NB: Negatif Big. The membership
degree of interactions and influences. The strength functions are considered as triangular for
AC/DC system is defined by the relative term Short (PM,PS,Z,NS and NM), and trapezoid for (PB and
Circuit Ratio (SCR). The (SCR) can usually be NB) as shown in Fig. 4.
expressed as the ratio of the power network short-
circuit SSC at the connection point from the station NB NM NS Z PS PM PB
with the alternative network to the continuous power
Pd of converters, either:

SCR  S SC Pd (2) Fig. 4. Membership function of inputs

With:
S SC  U 2 Z S (3) The initial values of the proportional and integral
gains (Kp0 and Ki0 ) of the conventional PI controller
are found using trial and error (tuned to best
Where SSC, U and ZS are short-circuit power of the performance). The error and rate of change of this
alternative network, the AC voltage between phases error are taken as inputs to the FL controller. These
and the impedance of the AC network at the inputs are normalized and then fuzzified using MFs.
fundamental frequency respectively. Then, these fuzzified inputs are applied to the rule
The weakness classification of AC/DC system as base for finding the output from the FL controller
classified using the short circuit ratio (SCR): weak ∆Kp and ∆Ki (or DeltaKp and DeltaKi) in Fig. 5,
systems are those having SCR between 3.0 and 2.0, when this outputs can be scaled using the scaling
whereas very weak systems have a value lower than factors (kp and ki which can be obtained by Genetic
2. When the SCR is greater than 3, the system is Algorithms) for obtain the best possible
strong.[11] performance. [3,12,13]
The rule base of outputs are summarized using
6. The Fuzzy Self-Tuning PI Controller Tables 1 and Table 2 respectively.
The Fuzzy Self-Tuning PI controller based on
fuzzy logic is combined with conventional PI
controller in both sides of converters, for updating
on line the values of the classical PI gains (Kp and
Ki) of current regulator and voltage regulator of
rectifier and inverter respectively. The errors for the
regulators and their derivatives are used as inputs to
the Fuzzy Self-Tuning PI Controller. The Fig. 3
shown the principle of its integration in HVDC
system.

Let
E  g ref  g meas (4)
R  ( E  E prev ) / Ts (5)
Fig. 5. Fuzzy Self-Tuning PI Controller in
Where E, Eprev, R, gref,gmeas and Ts are the error MATLAB/SIMULINK
 Tableau 1: Kp control rules 7. The STATCOM
“ERR RATE” The STATCOM is static synchronous generator
operated as a shunt-connected static var compe-
“ERR” NB NM NS Z PS PM PB
nsator whose capacitive or inductive output current
NB NB NB NB NB NB NM NS
NM NB NB NM NM NM NS Z
can be controlled independent of the AC system
NS NB NM NS NS NS Z Z voltage. The STATCOM based on a voltage-sourced
Z Z Z Z Z Z Z Z converter (VSC) is connected to the grid using a
PS Z Z PS PS PS PM PB coupling transformer at the inverter side of the
PM Z PS PM PM PM PB PB HVDC link and has a rating of ±100 MVA. It
PB PS PM PB PB PB PB PB consists of a three-level 48-pulse inverter and two
series-connected 3000 F capacitors which act as a
variable DC voltage source.
Tableau 2: Ki control rules The role of STATCOM is to exchange reactive
power with the network through a three-phase
“ERR RATE”
inductor, the latter being in general the leakage
“ERR” NB NM NS Z PS PM PB inductance of the coupling transformer.
NB NB NB NM NM NS Z Z The reactive energy exchange is done by
NM NB NB NS NS NS Z Z controlling the voltage of the inverter Vsh, which is
NS NB NM NS NS Z PS PS
NM NM NS Z PS PM PM
in phase with the busbar voltage Vr where the
Z
PS NS NS Z PS PS PM PB
STATCOM is connected (Fig 14).
PM Z Z PS PS PS PB PB Operation can be described as follows:
PB Z Z PS PM PM PB PB The flowage of active and reactive power
between these two voltage sources is given by:[8]
Defuzzification is carried out using the centroid
defuzzification method to obtain crisp values of ∆Kp VrVsh sin 
P (9)
and ∆Ki, where the center of gravity is used as final X sh
output. The output is:
V V  Vsh cos  
Q r r (10)
X sh
K 
  zi i
(6) From the equations (9) and (10), it can be seen
 i that when the two voltages are in phase ( = 0), there
Where i and zi are the membership grade of the is only a reactive power flows, the value of the
output membership function and the output variable, power exchanged depending only on the amplitude
respectively. of the two voltages Vr and Vsh.
The output of the FL controller is used to solve We can consider three possible cases always
the problem of fixed proportional and integral gains with ( = 0):
of a conventional PI controller (Fig. 6). The gains Kp If Vsh <Vr, the current in the inductance is
and Ki are updated online using the FL controller dephased by +pi/2 with to the voltage Vr and the
according to the following equations: current is capacitive.
If Vsh> Vr, the current in the inductance is
K p  K P 0  k p  K p (7) dephased by -pi/2 with to the voltage Vr and the
current is inductive.
K i  K i 0  ki  Ki (8) If Vsh = Vr, the current in the inductance is zero,
there is no exchange of energy.
Where Kp0 , Ki0, ∆Kp, ∆Ki, kp, ki Initial proportional
gain , initial integral gain, output of proportional 8. SIMULATION RESULTS
gain , output of integral gain, scaling factor from In order to study the performance of the HVDC
output of proportional gain and scaling factor from system, two cases of fault are realized with Fuzzy
output of integral gain, respectively. [14-15] Self-Tuning PI Controller and STATCOM, many
figures of simulations are presented in Figs.7 to 14

Fig. 6. Integration of FSTPIC in conventional PI controller of HVDC system

according to cases of table 3. 2.5

2
Current response at inverter side (whithout FSTPIC and whith STATCOM)

Current IdandIdref (pu)

1.5 1.3324 pu

Table 3: Different cases of simulation 1

conventional
0.5

FSTPIC STATCOM 0

Case PI -0.5
0 0.5 1 1.5 2 2.5 3

1 ×
Time (sec)

2 × Fig. 9. Single phase-to-ground fault at inverter side

3 × × (without FSTPIC and with STATCOM)
4 × × Current response at inverter side (whith FSTPIC and whith STATCOM)
2.5

Current IdandIdref (pu)

2

1) Single phase-to-ground fault at inverter side 1.5

1.2758 pu

The results of the simulation are presented in 1

Figs.7 to 10, in order to study the performance of the

0.5

HVDC system with proposed device. -0.5

0 0.5 1 1.5
Time (sec)
2 2.5 3

For a single phase to ground fault of 5 cycles, Fig. 10. Single phase-to-ground fault at inverter side
with breaker resistance equal 40, the HVDC (with FSTPIC and with STATCOM)
system with (FSTPIC) controller and STATCOM
gives good transient performance in terms of lower 2) Three phase-to-ground fault in inverter side
peaks ( 1.2758 pu in Fig.10 ) and a better shape For a three phase-to-ground fault of 5 cycles,
when compared with the others cases. The settling between t=1.2s and t=1.3s the HVDC system with
time is very low in the last case (Table 4 and 5) in (FSTPIC) and STATCOM gives good transient
rectifier and inverter current response. performance in terms of lower peaks ( 1.5178 pu in
Fig.14) and a better shape.
Table 4: Performance comparison of single phase to In the starting of the system in note that the
ground fault case in rectifier side. (FSTPIC) imposes an influence on the performances
Overshoot Rise Settling of the system that it translates by a good shape of the
Devices With current Id in the curves when the (FSTPIC) is
Peak (pu) Time Time
HVDC system applied (in Fig.12 and 14)
(s) (s)
Conventional PI 1.690 0.774 0.96 The FSTPIC can better withstand the oscillations
of the current and leads to a rapid response, which
FSTPIC 1.286 0.487 0.2 restore the system more quickly to stability. The
STATCOM 1.167 0.889 0.35 settling time after fault is very low in the cases when
FSTPIC+STATCOM 1.079 0.505 0.135 the (FSTPIC) is applied.
In cases where the STATCOM is applied notice
Table 5: Performance comparison of single phase to that undershoot of the current does not exceed the
ground fault case in inverter side. value of 0.5 pu when the existence of the default.
Overshoot Rise Settling After connecting STATCOM, the HVDC system
Devices With with STATCOM gives a good waveform and
Peak (pu) Time Time
HVDC system
(s) (s) mitigates the current. In this, the value of amplitude
Conventional PI 1.91 0.7673 0.8 (in Fig.13 equal 1.5355 pu) is less than the case
FSTPIC 1.779 0.482 0.193 where the STATCOM is not connected (in fig.11 the
STATCOM 1.332 0.881 0.285 peak is 2.5868pu). That translates a more stable
FSTPIC+STATCOM 1.276 0.492 0.129 performance.
The settling time is very low in last case (Table 6
and 7) in rectifier and inverter current response.
Current response at inverter side (whithout FSTPIC and whithout STATCOM)
2.5

1.9052 pu
Current IdandIdref (pu)

2 Current response at inverter side (whithout FSTPIC and whithout STATCOM)

3
2.5868 pu
1.5 2.5
f (pu)

2
ndIdre

1
1.5
0.5
t Ida

1
rren

0 0.5
Cu

0
-0.5
0 0.5 1 1.5 2 2.5 3 -0.5
Time (sec) 0 0.5 1 1.5 2 2.5 3
Time (sec)

Fig. 7. Single phase-to-ground fault at inverter side Fig. 11. Three phase-to-ground fault at inverter side
(without FSTPIC and without STATCOM) (without FSTPIC and without STATCOM)
Current response at inverter side (whith FSTPIC and whithout STATCOM)
2.5 Current response at inverter side (whith FSTPIC and whithout STATCOM)
3

2
Current IdandIdref (pu)

2.5
u)

1.7787 pu
f (p

2 1.9427 pu
dIdre

1.5
1.5
t Idan

1 1
rren

0.5 0.5
Cu

0
0
-0.5
0 0.5 1 1.5 2 2.5 3
-0.5 Time (sec)
0 0.5 1 1.5 2 2.5 3

Fig. 12. Three phase-to-ground fault at inverter side

Time (sec)

Fig. 8. Single phase-to-ground fault at inverter side (with FSTPIC and without STATCOM)
(with FSTPIC and without STATCOM)
 3. Aurobinda, R., Dash, P.K., Panda, S.K.: A Fuzzy Self-
Current response at inverter side (whithout FSTPIC and whith STATCOM)
3

2.5
dIdref (pu)

1.5
1.5355 pu
Tuning PI Controller for HVDC Links, in: IEEE
Transactions on Power Electronics, Vol.11, NO.5, pp.
t Idan

1
rren

0.5

669-679, (1996).
Cu

-0.5
0 0.5 1 1.5 2 2.5 3
Time (sec)
4. Yoon, J.Y., Hwang, G.H., Park, J.H.: A Genetic
Fig. 13. Three phase-to-ground fault at inverter side Algorithm Approach to Design an Optimal Fuzzy
(without FSTPIC and with STATCOM) Controller for Rectifier Current Control in HVDC
System, in: The 1998 IEEE International Conference on
Evolutionary Computation Proceedings, vol., no., pp.404-
Current response at inverter side (whith FSTPIC and whith STATCOM)
3

2.5

409,( 1998).
f (pu)

2
dIdre

1.5178 pu
1.5

5. Shahgholian, G., Mahdavian, M., Emami, A.,

t Idan

1
urren

Ahmadzade, B.: Improve power quality using static

0.5
C

-0.5
0 0.5 1 1.5
Time (sec)
2 2.5 3
synchronous compensator with fuzzy logic controller, in:
Fig. 14. Three phase-to-ground fault at inverter side International Conference of Electrical Machines and
(with FSTPIC and with STATCOM) Systems (ICEMS), pp. 1-5, (2011).
6. M. Ramesh , M., Laxmi, J. A, Power Transfer
Table 6: Performance comparison of three phase to Capability Improvement to HVDC Transmission System
ground fault case in rectifier side. using Artificial Neural Network and Inference System
(ANFIS) Controller, in: Asian Power Electronics Journal,
Overshoot Rise Settling
Devices With Vol. 8, No. 3,( 2014).
Peak (pu) Time Time
HVDC system 7. Meah, K., Ula, S.: Investigation on fuzzy logic based
(s) (s)
auto-tuning current controller application in HVDC
Conventional PI 2.037 0.773 0.399
Links. In: IEEE region 5 conference. AR, USA:
FSTPIC 1.356 0.486 0.258
University of Arkansas; pp. 266–272, (2007).
STATCOM 1.332 0.895 0.306
FSTPIC+STATCOM 1.197 0.505 0.225 8. Khatir, M., Zidi, S.A., Fellah, M.K., Hadjeri, S., Flitti,
M.: The Impact Study of a Statcom on Commutation
Table 7: Performance comparison of three phase to failures in an HVDC Inverter Feeding a Weak AC
ground fault case in inverter side. System. in: Journal of Electrical engineering, Vol. 63,
NO. 2, pp. 95–102, (2012).
Overshoot Rise Settling
Devices With 9. Paulinder, J.: Operation and Control of HVDC Links
Peak (pu) Time Time
HVDC system Embedded in AC Systems, Licentiate thesis of
(s) (s)
Conventional PI 2.587 0.768 0.5 engineering, Chalmers University of Technology,
FSTPIC 1.943 0.48 0.28 Gôteborg, Sweden,( 2003).
STATCOM 1.536 0.88 0.42 10. Arrillaga, J.: High voltage Direct Current
FSTPIC+STATCOM 1.518 0.49 0.22 Transmission, IEE Power engineering series 29. 2nd
Edition, Printed in England by short Run press Ltd,
Exeter, UK.(1998).
9. Conclusion
In this article, a method combining FSTPIC with 11. Sood, V.K.: HVDC and FACTS Controllers
a STATCOM was integrated in the HVDC system. Applications of Static Converters in Power Systems,
Printed by Kluwer Academic Publishers, Boston, (2004).
A performance comparison between four cases 12. Khatir, M., Zidi, S.A., Hadjeri, S., ., Fellah, M.K.:
includes different controls showed that the
Analysis of Recovery From Commutation Failures in an
robustness and the adaptation of the proposed device
HVDC Inverter Connected to a Weak Receiving AC
is very efficient. When the AC system is weak, the
System, in: Acta Electrotechnica et Informatica Vol. 8,
HVDC system is sulfur with the conventional
No. 1, pp. 44–50, (2008).
controller while the proposed controller has
satisfactory performance and consistently performs 13. Qi, J., Sood, V.K., Ramachandran, V.: Incremental
better under all conditions such as low settling time Fuzzy PI Control of a HVDC Plant, in: Proceedings of the
after fault, low peak and no occur of commutation 2005 IEEE Conference on Control Applications, Toronto,
failure in many cases. Canada,pp.1305-1310, ( 2005).
14. Meah, K., Ula, S.: Simple Fuzzy Self-tuning PI
References. Controller for Multi-terminal HVDC Transmission
1. Joncquel, E. : transport d’énergie en courant continu à Systems, In: Electric Power Components and Systems,
haute tension (Power Transmission of High Voltage 36:pp.224–238, (2008).
Direct Current), in :Techniques de l’Ingénieur, D 4761.
15. Munish, M.: Design & Optimization of Intelligent PI
2. Atighechi, H.,Chiniforoosh, S., Jatskevich, J. and al.:
Dynamic average-value modeling of CIGRE HVDC Controllers (Fuzzy & Neuro-Fuzzy) for HVDC
benchmark system, in: IEEE Trans. Power Delivery, vol. Transmission System, Master of Applied Science Thesis,
29, NO.5, pp. 2046-2054, (2014). University of Ontario, Canada, (2010)