See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/353106032

SRF Control Strategy based on STFIS Controllers For Single- Stage Four-Leg
Transformerless Inverter for Grid-Connected PV system

Article  in  Przeglad Elektrotechniczny · July 2021

CITATIONS READS

0 14

6 authors, including:

Djamel Eddine Chaouch Hocine Abdelhak Azzeddine

University Mustapha Stambouli of Mascara University Mustapha Stambouli of Mascara
19 PUBLICATIONS   53 CITATIONS    24 PUBLICATIONS   6 CITATIONS   

SEE PROFILE SEE PROFILE

Zoubir Ahmed Foitih

Université des Sciences et de la Technologie d'Oran Mohamed Boudiaf
49 PUBLICATIONS   199 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Développement d'un logiciel de simulation des systèmes PV View project

Aerial Manipulation View project

All content following this page was uploaded by Hocine Abdelhak Azzeddine on 09 July 2021.

The user has requested enhancement of the downloaded file.

 1. Djamel Eddine CHAOUCH 1, 2. Hocine Abdelhak AZZEDDINE1, 3. Ahmed LARBAOUI1,
4. Zoubir AHMED-FOITIH2
University Mustapha Stambouli of Mascara, Algeria (1), University of Sciences and Technology of Oran – Mohamed Boudiaf Algeria (2)
ORCID: 1. 0000-0002-5130-4229; 3. 0000-0001-7520-9093; 4. 0000-0003-3121-9964

doi:10.15199/48.2021.07.08

SRF Control Strategy based on STFIS Controllers For Single-

Stage Four-Leg Transformerless Inverter for Grid-Connected PV
system
Abstract. In this paper synchronous reference frame (SRF) control strategy based on a self-tunable fuzzy inference system (STFIS) is proposed to
control a four-leg transformerless inverter, is presented. The single-stage inverter is controlled to achieve high performance from the PV system
without using any maximum power point technique (MPPT), to inject power into the grid, and to improve the power quality, simultaneously. A Fuzzy
controller based on an on-line optimization of a zero-order Takagi-Sugeno fuzzy inference system (FIS) by a back propagation-like algorithm is
applied to minimize a cost function that is made up of a quadratic error term and a weight decay term that prevents excessive growth of DC-link and
PCC. A detailed design of the STFIS controller parameters is given in detail. Furthermore, a comparison between classical PI and STFIS controllers
is provided. The performance test of the proposed single-phase configuration and its control strategies are realized under the presence of severe
conditions using Matlab/Simulink.

Streszczenie. W artykule przedstawiono strategię sterowania synchroniczną ramką odniesienia (SRF) w oparciu o samostrajalny system
wnioskowania rozmytego (STFIS) do sterowania falownikiem beztransformatorowym. Falownik jednostopniowy jest sterowany w celu uzyskania
wysokiej wydajności systemu fotowoltaicznego bez stosowania techniki maksymalnego punktu mocy (MPPT), jednoczesnego wprowadzania mocy
do sieci i poprawy jakości energii. Sterownik Fuzzy oparty na optymalizacji on-line systemu wnioskowania rozmytego Takagi-Sugeno (FIS)
zerowego rzędu za pomocą algorytmu przypominającego propagację wsteczną jest stosowany w celu zminimalizowania funkcji kosztu, która składa
się z kwadratowego składnika błędu i wagi zaniku, który zapobiega nadmiernemu wzrostowi łącza DC i PCC. Szczegółowo podano projekt
parametrów sterownika STFIS. Ponadto przedstawiono porównanie między klasycznymi sterownikami PI i STFIS. (Strategia sterowania SRF
oparta na sterownikach STFIS dla jednostopniowego falownika beztransformatorowego dla systemu fotowoltaicznego podłączonego do
sieci)
Keywords: SRF control, Four-Legs transformerless converter, single-stage converter, STFIS controller, power quality improvement.
Słowa kluczowe: falownik, sterownik STFIS, poprawa jakości energii

Introduction under the presence of selected conditions. Unfortunately,

In the last decade, the power energy demand is rapidly this approach is not tested under sudden variation of solar
increasing, due to the population and economic growth, irradiation and shading. Also, the use of simple PI cannot
especially for large countries. To satisfy this demand, many perform well under sudden variations, which causes
countries in the world adopted a higher penetration of divergence of the control.
renewable energies, such as wind turbines and solar Many control strategies are proposed in the literature to
photovoltaic systems in their utilities. Compared to wind control the interfacing inverter, such as PQ theory-based
turbines, solar photovoltaic systems (PVs) can easily install control, nonlinear control, indirect and direct control, and
on the roofs of houses and factories, which facilities SRF control, etc. [10]. Most of the proposed control
integration of the generated power into the grid and strategies detailed in [10], employed classical PI controllers,
supplied load locally. which show poor control performances for an integrating
According, to [1], there are many configurations to process and large time delay process and can easily be
connect solar photovoltaics to the grid. However, one can saturated during sudden variation. Also, is very sensitive to
use either a single-stage inverter or a two-stage inverter to the noise caused by nonlinear loads [11]. So gain tuning
connect the PVs to the grid [2, 3]. According to the and additional action such as integration of the anti-windup
comparison study between the existing configurations for feedback to avoid saturation issue are suggested in [12,
PVs connected to the grid realized by [4], a single-stage 13]. For better performance, authors in [14], proposed
configuration is a better view of point efficiency, and for the proportional resonant controllers with at-windup feedbacks
DC voltage stability, a two-stage configuration is more for the inner and outer control loops to the voltage and
stable, which leads that the power quality improvement can frequency control for the standalone system. The obtained
be achieved correctly using this configuration. Compared to results show satisfactory performance under load variation
the two-stage converter configuration, single-stage and noise conditions. In [15], an adaptive neuro-fuzzy
converter employe only one converter and less sensors as interference system (ANFIS) controller is suggested for two
detailed in [5, 6]. However, an accurate passive element stage-converter configurations. Regarding, the dynamic
and controller gain design is required to achieve high response, saturation, and noise issue, the controllers
performance. Also, to get the maximum power point suggested in [14-15], perform well, but they are more
tracking (MPPT) from the PVs, a selected MPPT technique complex and require more computation time.
is required [7]. Generally, the maximum PV voltage should This paper introduces the single-stage configuration
be equal to the reference DC link voltage, which is hard to transformerless inverter for the grid-connected-PV system,
realize due to the sudden solar irradiation change or and synchronous reference frame (SRF) based control with
shading. This objective is solved in [8], using Lyapunov- a self-tunable fuzzy interface system to achieve the
based control. This approach is effective but more complex. following objectives as, 1)Less power losses due to power
To reduce the control complexity and its implementation in conversion, 2)Injection of the generated power from PVs
real-time, active and reactive control based on the without power quality issues, 3)Stability of DC link and AC
proportional-integral controller (PI) is proposed in [9]. The voltages in the presence of severe conditions, 4)High PV
obtained results in [9] show a satisfactory performance performance without MPPT technique, 5)Optimal STFIS

42 PRZEGLĄD ELEKTROTECHNICZNY, ISSN 0033-2097, R. 97 NR 7/2021

controller parameters design to achieve high performance achieving MPPT by controlling the DC voltage, 2) improving
during the transition and in presence of charting signal the power quality synchronizing with PCC. RC ripple filter is
problems. connected between the four-leg converter and grid to
eliminate the ripple voltage due to high-frequency switching.
System understudy
A nonlinear load is connected to the PCC. To allows a
Figure.1 shows the single-stage configuration
three-phase system to use high voltage while still
transformerless inverter for the grid-connected-PV system
supporting lower voltage single phase application, a neutral
under study. It consists of photovoltaic panels connected in
wire is used.
series with total output voltage closer to the DC-link voltage
connected to the grid through a four-leg two-level converter
which is controlled to achieve many tasks, such as, 1)

Fig.1. Single-stage four-leg transformerless configuration for grid-connected PV system

Figure.2 demonstrates the characteristics VPV=f(IPV) of  1

the selected photovoltaic panels, which are connected in  cos   sin  
series to get an output voltage closer to the DC link voltage, iLd   2 i
 La 
which is equal to 350 V. As shown in Fig.2 regulating the (1)   2  2   2  1  
DC-link voltage at 350 V, one can easily extract the MPPT.  iLq   3  cos    3  sin   
3
  iLb
2  
i      
 iLc 
This value represents the maximum voltage Vmp that  Lo    2   2  1
extracts if using the conventional MPPT techniques. So,  cos     sin     
using this concept one can reduce the control system   3   3  2
complexity.
where: iLd, iLq, and iLo represent the load currents in the dqo
axis.

Fig.2. Characteristics VPV=f(PPV) and VPV=f(IPV) under solar Fig.3. SRF-based control with STFIS controllers
irradiation and fixed temperature
A three-phased locked loop (PLL) is employed to synchronize
Table 1. PV voltage, solar irradiation, and current the signals with grid voltages by calculating (cosϴ,sinϴ). The d-q
VPV[V] G [W/m2] IPV [A] load current components are passed through a low pass filter (LPF)
350 300 0.3 to extract the DC components (iLdc and iLqc). The reference of
* *
350 500 1.33 source currents (iGd , iGq ) in d-q axis are equal to;
350 600 1.95
350 700 2.7 iGd *  iLdc  iloss
350 800 3.65 (2)  *
350 900 4.65  iGq  iqV  iLqc
350 1000 5.8
Control system where: iLdc, iLqc, iloss, ilqc represent the dq-axis component of
In Figure.3 the enhanced SRF based control based on the load current; where:iloss, ilqc represent the active and
STFIS controllers proposed in [10] is presented. The load reactive power components, which are obtained at the
currents (iLa, iLb, iLc), grid voltages (vGa, vGb, vGc), grid current output of the STFIS controllers of the DC-link and AC
(iGa, iGb, iGc), DC link voltage (Vdc), and neutral current (iN), of voltages.
the single-stage converter, are sensed to use as feedbacks With help of reverse Park’s transformation of the
signals. With help of the Park’s the load currents in the equation in (2), one obtains the references of grid currents
three phases are transformed into dqo frame as, (iGa*, iGb*, iGc*) as,

PRZEGLĄD ELEKTROTECHNICZNY, ISSN 0033-2097, R. 97 NR 7/2021 43

   parameter,  i n represents derived term of the cost function
 cos  sin  1 i * 
iGa 
*
  Gd ( i neuron of layer n),  j
th n 1
is the output of j
th
unit of
(3)  * 2   2   2   
iGb    cos     sin     1  iGq*  layer (n-1), y1 is the output value, and y represent the
 * 3   3   3   *
iGc    2   2    iGo  dedesired output.
cos     sin     1 As demonstrated in Fig.4, The Membership functions are
  3   3   normalized and shared in five subsets for all displacement.
* * *
The reference grid currents (iGa , iGb , iGc , iN ) and the
* The linguistic labels are defined as follows: NB: Negative
neutral reference current with respective sensed currents Big, NS: Negative Small, Z approximately Zero, PS: positive
(iGa, iGb, iGc, iN) are fed to current-controlled hysteresis Small, and PB: Positive Big. The STFIS network
controllers to generate the switching signals to (S1-S8). parameters for both outer control loops consist of five
memberships functions of type sigmoid and Gaussians,
STFIS controllers design where their parameters are equal to   0.0001 ,   0.3
As detailed in Fig.2, both outer control loops employed self-
tuning fuzzy inference system (STFIS) controllers, which and b  0.9 .
are a neuro-fuzzy network that provides a complete
structural system with the fuzzy inference of zero-order type
Takagi-Sugeno. Its architecture, which consists of four
layers as detailed in [16, 17]. However, the errors in
voltage (∆V and ∆Vdc ) of the AC and DC voltages are
received by the first layer, and the membership degrees of
these errors are calculated in layer two. The value of the
truth is calculated in the third layer. The fourth layer is
designed to provide the outputs, it calculates the values of
and the outputs iqv, and iloss. To adjustment of the
backpropagation, the gradient method is used. The
optimization is conducted entirely online by minimizing a
cost function J to generate the parameters wi, Fig.4. Memberships function
characterizing the conclusion part of rules, and adjust them.
However, The gradient descent algorithm with the
regression parameters optimizing only the conclusion part
of rules is adapted to meet the targets. The cost function is
expressed as,
(4) J  E w 2

where: w ,  , and E denote the weight, constant that

controls the growth parameters, and the quadratic error,
respectively.
With help of the backpropagation method and the effect of
the regression term of the weights in the cost function, the
parameters are adjusted as,
J
(5) w(t  1)  w(t )   ( )  b w(t )
w Fig.5. The parameters (weights)
With the help of Eq.4 and taking the regression coefficient
Figure.5 represents the convergence of the parameters
 equal to 2 , Eq.5 is rewritten as, (weights) of the four-layer. One observes clearly that the
E STFIS controllers achieve a good tracking feature during
(6) w(t  1)  w(t )   ( )  b w(t )   w(t ) the transition and at steady-state.
w
Equation (6) is updated by multiplying the regression Results and discussion
coefficient by  i /  i i To test the performance of the single-stage
transformerless four-leg converter grid-connected PV
where  i is the truth value of the premise part of the system and the improved enhanced SRF based-control with
triggered rule. STFIS controllers, the system is subjected to sudden solar
However, one can easily incorporate the effect of the irradiation and sudden balanced and unbalanced nonlinear
regression terms in the cost function, which is limited for the load variation.
optimization of the output as, In Figure.6 the dynamic performance of the grid
(7) w14j (t )  14 3j  bw14j (t  1)   3j  w 14j (t  1) /  3
j
voltages (VG), grid currents (iG), load currents (iL), inverter
currents (iinv), output PVs current (iPV), DC-link voltage (Vdc),
and *
and its reference (Vdc ), and neutral current (iN), are
(8) 14  y1  y /  j
3
j
demonstrated. System is subjected to sudden solar
irradiation at t= 0.4 s, t=0.9s, t=1.4s, t=1.9 s, and at t=2.4s.
Where wijn (t ) : i th is the parameter between i of layer n Also, the system is subjected to the load change at t=1.9s.
one observes that DC In link voltage flows it reference
and j st unit of layer ( n  1 ),  i ,  is the positive learning during solar irradiation, also the AC voltage is maintained
constant during these sudden variations, which confirms the
gain, t is the training iteration, b is the moment robustness of the STFIS controllers for the outer loops

44 PRZEGLĄD ELEKTROTECHNICZNY, ISSN 0033-2097, R. 97 NR 7/2021

control. One observes clearly that at t=1.9s the unbalanced
nonlinear load is suddenly connected to the PCC that is
why the neutral current appears at this time. To balance the
grid current due to the unbalance created by the
unbalanced nonlinear load, the single-stage four-leg-
converter acts as an active filter. The maximum output PVs
currents at (0.4s, 0.9s, 1.4s, 1.9s, and 2.4s) are equals to
(0.3A, 1.95A, 3.65A, 5.8A, and 1.95A) for solar irradiations
2 2 2 2 2
(300 W/m , 600W/m , 800W/m , 100W/m , and 600W/m ),
which is following the obtained results shown in Fig.2 and
detailed in Table.1, which confirms the proposed approach
to extract the maximum of currents from PVs with
controlling only the DC-link voltage without using any
MPPTs technique.

Fig.8. Zoom of the dynamic performance under load and solar

irradiation change between t=1.8s and t=2s

In Figure.7 the zoomed waveforms between t=1s to

t=1.5s of the performance shown in Fig.4, are presented.
One observes clearly that the grid currents are sinusoidal
instead of the presence of balanced nonlinear load. One
sees that at t= 1.4s, the grid currents are decreased due to
increasing of the output PV current. The DC link voltage at
the sudden variation of solar irradiation from 600W/m2 to
800W/m2 at t=1.4s is not affected and the overshoot does
not exceed 1V, which confirms the robustness of the STFIS
regulator to maintain stable the DC link voltage at the
sudden variation.
Fig.6. Dynamic performance under load and solar irradiation In Figure.8 the zoomed waveforms between t=1.8 s and
change t=2 s of the performance shown in Fig.4, are presented. It is
observed at t=1.9 s the neutral current appears because of
the unbalance in load current. Instead of this perturbation,
the grid current is balanced and sinusoidal. The DC link
voltage and AC voltage are well regulated in this period.
Also, the PVs current increases from 3.65 A to 5.8 A due to
an increase of the solar irradiation from 800 W/m2 to 1000
2
W/m . This confirms the good operation of the enhanced
SRF control based STFIS controllers in the presence of
unbalanced nonlinear load and sudden variation of solar
irradiation.

Fig.9. Grid, PVs, and load active powers during variation of load
and solar irradiations
Fig.7. Zoom of the dynamic performance under load and solar
irradiation change between t=1s and t=1.5s In Figure.9 the waveforms of the active powers of load
(PLoad), grid (grid), and PVs (PPV) are demonstrated. One
observes that the grid power varies with the variation of the

PRZEGLĄD ELEKTROTECHNICZNY, ISSN 0033-2097, R. 97 NR 7/2021 45

output PV currents, It decreases with the increase of the choice for this application that necessitates a stable DC link
generated power from the PVs. One observes that load is voltage to extract the MPPT from the PVs. One observes in
supplied without interruption at all times, which confirms the Fig.11 (b) the THD of the grid current when to use PI
good operation of the proposed configuration and their controllers are 4.99%, however, in Fig.11 (c) the THD of
control strategies based on STFIS controllers. the grid current when using the STFIS controller is 1.36%.
In Figure.10 the harmonic spectrum of a) load current, b) Both THD respects the recommendation of IEEE-519-1992,
grid current, and c) the grid voltage, are presented. One but that with STFIS controller is perfect.
observes that the total harmonic distortion of grid current
and voltage current is less than 5%, which respects the
recommendation harmonics levels in IEEE Std 519-1992.

(a)

(a)

(b)

(b) (c)
Fig. 11. Harmonic spectrum of a) load current, b) grid current, and
c) grid voltage

Conclusion
In this paper single-stage transformerless four-leg
(c) converter configuration for grid-connected PVs is selected
Fig. 10. Harmonic spectrum of a) load current, b) grid current, and for low voltage application. Enhanced synchronous
c) grid voltage reference (SRF) based control with STFIS controllers for
DC link voltage and AC voltage regulators, is selected and
In Figure.11 (a) the waveforms of the DC link voltage implemented. An accurate gain design for the STFIS
regulation using the STFIS controller and the conventional controllers and comparison study SFIS and PI controllers
PI controller is presented. One observes clearly that during are provided. It has been demonstrating the capability of
the sudden variation of the solar irradiation STFIS controller these advanced controllers to maintain stability under the
is more stable than the PI controller, which confirms our

46 PRZEGLĄD ELEKTROTECHNICZNY, ISSN 0033-2097, R. 97 NR 7/2021

presence of noise in the AC and DC voltages. It has been connected photovoltaic systems." IEEE Transactions on
confirmed that one can extract the MPPT from PVs by Energy Conversion 26, no. 2 (2011): 707-715.
controlling the DC-link voltage and without using any MPPT [7] Jain, Sandeep, and Vivek Agarwal. "Comparison of the
performance of maximum power point tracking schemes
methods. The obtained results viewpoint THD and response
applied to single-stage grid-connected photovoltaic systems."
time under the presence of severe conditions such as solar IET Electric Power Applications 1, no. 5 (2007): 753-762.
irradiation and unbalanced nonlinear load are satisfactory. [8] Meza, Carlos, Domingo Biel, Dimitri Jeltsema, and Jacquelien
MA Scherpen. "Lyapunov-based control scheme for single-
Authors: dr. Djamel-Eddine Chaouch, Faculty of Science and phase grid-connected PV central inverters." IEEE Transactions
Technology, University Mustapha Stambouli of Mascara, Algeria E- on control systems technology 20, no. 2 (2011): 520-529.
mail: [email protected]; dr. Hocine Abdelhak [9] Sreekanth, Thamballa, Narasimhamurthy Lakshminarasamma,
Azzeddine, LSTE Laboratory, Faculty of Science and Technology, and Mahesh K. Mishra. "Grid tied single-stage inverter for low-
University Mustapha Stambouli of Mascara, Algeria E-mail: voltage PV systems with reactive power control." IET Power
[email protected]; dr. Ahmed LARBAOUI, Electronics 11, no. 11 (2018): 1766-1773.
Faculty of Science and Technology, University Mustapha Stambouli [10] Singh, Bhim, Ambrish Chandra, and Kamal Al-Haddad. Power
of Mascara, Algeria E-mail: [email protected]; prof. quality: problems and mitigation techniques. John Wiley &
Zoubir Ahmed-Foitih, Faculty of Electrical Engineering, University Sons, 2014.
of Sciences and Technology of Oran, El M‟naouer, Oran, Algeria, [11] Yang, Ming, Dian-guo Xu, and Xian-guo Gui. "Review of
E-mail: [email protected]. control system anti-windup design." Electric Machines and
Control 10, no. 6 (2006): 622.
REFERENCES [12] Wang, Liuping. "Automatic tuning of PID controllers using
[1] Ueckerdt, Falko, Robert Brecha, and Gunnar Luderer. frequency sampling filters." IET Control Theory & Applications
"Analyzing major challenges of wind and solar variability in 11, no. 7 (2017): 985-995.
power systems." Renewable energy 81 (2015): 1-10 [13] Rezkallah, Miloud, Abdelhamid Hamadi, Ambrish Chandra,
[2] Islam, Md Rabiul, A. M. Mahfuz-Ur-Rahman, Kashem M. and Bhim Singh. "Design and implementation of active power
Muttaqi, and Danny Sutanto. "State-of-the-art of the medium- control with improved P&O method for wind-PV-battery-based
voltage power converter technologies for grid integration of standalone generation system." IEEE Transactions on
solar photovoltaic power plants." IEEE Transactions on Energy Industrial Electronics 65, no. 7 (2017): 5590-5600.
Conversion 34, no. 1 (2018): 372-384 [14]Rezkallah, Miloud, Ambrish Chandra, Marco Tremblay, and
[3] Wu, Tsai-Fu, Chih-Hao Chang, Li-Chiun Lin, and Chia-Ling Hussein Ibrahim. "Experimental implementation of an APC with
Kuo. "Power loss comparison of single-and two-stage grid- enhanced MPPT for standalone solar photovoltaic based water
connected photovoltaic systems." IEEE Transactions on pumping station." IEEE Transactions on Sustainable Energy
Energy Conversion 26, no. 2 (2011): 707-715 10, no. 1 (2018): 181-191.
[4] Zhu, Yongli, JianGuo Yao, and Di Wu. "Comparative study of [15]Benhalima, S., Chandra, A. and Rezkallah, M., 2019. Real-time
two stages and single stage topologies for grid-tie photovoltaic experimental implementation of an LMS-adaline-based ANFIS
generation by PSCAD/EMTDC." In 2011 international controller to drive PV interfacing power system. IET Renewable
conference on advanced power system automation and Power Generation, 13(7), pp.1142-1152.
protection, vol. 2, pp. 1304-1309. IEEE, 2011. [16]H. Maaref, C. Barret, “Progressive Optimization of a Fuzzy
[5] Patel, Hiren, and Vivek Agarwal. "MPPT scheme for a PV-fed Inference System,’’ IFSA-NAFIPS’2001, Vancouver, pp.665-
single-phase single-stage grid-connected inverter operating in 679, 2001.
CCM with only one current sensor." IEEE Transactions on [17]K. Zemalache, H. Maaref “Intelligent control for a drone by
Energy Conversion 24, no. 1 (2009): 256-263. self-tunable fuzzy inference system,” 6th international multi-
[6] Wu, Tsai-Fu, Chih-Hao Chang, Li-Chiun Lin, and Chia-Ling conference on systems, signals and devises, Djerba, Tunisia,
Kuo. "Power loss comparison of single-and two-stage grid- 23-26 March 2009.

PRZEGLĄD ELEKTROTECHNICZNY, ISSN 0033-2097, R. 97 NR 7/2021 47

View publication stats