Journal of Control, Automation and Electrical Systems

https://doi.org/10.1007/s40313-023-00992-9

Control Design of Single-Phase T-Type Inverters for PV Residential

Applications: A Practical Approach
Manh Linh Nguyen1 · Phuong Vu1

Received: 5 August 2022 / Revised: 12 February 2023 / Accepted: 15 February 2023

© Brazilian Society for Automatics–SBA 2023

Abstract
In this research, a practical solution is proposed to enhance the performance of the single-phase DC/AC converter, which is
usually used as an interface between the renewable energy source and the power grid in residential applications. In order to
meet the strict requirements of the grid code, various solutions have been applied. In detail, the multilevel T-type topology
is employed to further reduce the distortion of the output current. In control design, adaptive sliding mode control designed
in discrete-time domain is adopted to enhance the robustness of the inner current control-loop against parameter variations.
For the dc bus voltage control loop, multiple objectives such as negligible overshoot and limited inrush current during the
startup process, and reduced voltage oscillation with the sudden change of the transferred power, can be fulfilled by using
state-machine programming and Fuzzy-PI controller. The effectiveness of the proposed solution is verified by numerical
simulations where all control algorithms are implemented in C-language. This technique guarantees consistency between
simulations and practical implementations based on digital platforms.

Keywords T-type inverter · Grid connected · PV · Discrete-time control · Adaptive control

1 Introduction erated by the PV to an appropriate level to be used by the

second converter. The second stage converter always uses
Renewable energy has been proven as the major solution, to DC/AC topology, which has the ability to regulate the DC
reduce not only the dependence on conventional fossil ener- link voltage, as well as bidirectional power-flow control. With
gies but also the negative impacts on the environment. Over the explosive development of residential rooftop PV sys-
the last two decades, the worldwide cumulative installed tems, single-phase grid-connected converters have become
capacity of renewable energy sources (RES) such as pho- an interest of the solar market. Hence, this research focuses
tovoltaic (PV), wind, ocean, etc., has grown significantly. on improving the performance of this type of converter, in
Those RES are not only installed on large scales such as terms of robustness against parameter variations and the per-
solar and wind farms, but also on small scales, i.e., residen- formance of the grid-connected current.
tial rooftop PV systems. Currently, commercial converters manufactured by large
To effectively collect solar energy as well as transfer it to companies such as REFU, SMA, Energy, Danfoss Solar, and
the grid, PV and two-stage converters are usually employed. Subways have been already available on the market, with
A DC/DC converter that is capable of performing maximum maximum efficiency up to 98%. The most popular topologies
power point tracking (MPPT) is adopted as the first stage. used for grid-connected converters are H-bridge and neutral
This DC/DC converter also transforms the DC voltage gen- point clamped (NPC) (Kjaer et al., 2005). The main reason is
that those above-mentioned topologies exhibit bi-directional
power-transfer ability, which is obligatory in small-capacity
B Manh Linh Nguyen
PV systems due to the existence of backup batteries (Sulae-
[email protected]
man et al., 2016). However, the quality of a grid-connected
B Phuong Vu
converter is ensured not only by its topology but also by its
[email protected]
control strategy.
1 School of Electrical and Electronic Engineering, Hanoi To guarantee the stable and secure operation of the power
University of Science and Technology, No.1A, Dai Co Viet, system network, strict requirements of the grid code such as
Hanoi 10000, Vietnam

123
 Journal of Control, Automation and Electrical Systems

variation in voltage and frequency range, harmonic currents By considering all the above-mentioned issues, various
(order-h) limits, maximum current total harmonic distortion solutions have been proposed in this research to improve the
(THD), DC current injection (Tsili & Papathanassiou, 2009), performance of the single-phase grid-tied converter. First,
must be met by each of grid-connected converters. Remark- adaptive sliding mode control (ASMC) is adopted for the
able efforts focusing on control design have been carried out inductor current control loop. This approach not only gives
to assure that those grid-connected converters can meet the a quick transient response but also removes the chattering
grid-code requirements (Zeb et al., 2018). Conventionally, phenomenon since the switching control action is not nec-
the two cascaded loops control system is usually adopted, essary. In addition, the controller is also robust against the
including an outer DC bus voltage-loop, and an inner current- tolerance of inductance during operation caused by tempera-
loop. Particularly, the current controller design attracts great ture variation. Second, a combined solution where fuzzy-PI,
attention since it has a decisive influence on the transient and state-machine techniques are employed to guarantee that
response as well as the THD of the grid-connected cur- the overshoot of the DC-bus voltage during start-up is neg-
rent. Most commercial devices use voltage-oriented control ligible, the inrush current is restricted, whilst the transient
(VOC), where a synchronous rotating reference frame and response caused by load-variation in normal operation is
conventional PI are utilized. This approach is simple and improved. The validity of the proposed solution is confirmed
straightforward in control design as well as parameter tuning. by numerical simulations in which all control algorithms are
However, limited bandwidth and coupling between the direct implemented by standard C-language. This technique is very
and quadrature current are the main issues of this method. In important since it guarantees consistency between simula-
(Wang et al., 2014; Silwal et al., 2019; Guo & Guerrero, tions and practical implementations.
2016), the proportional-resonant (PR) controller designed in
static reference frame αβ is discussed. By utilizing the high
gain at its resonance frequency, this method not only can 2 System Description
minimize the coupling effect but also can improve the THD
of the output current by using PR with multiple resonant fre- The typical topology of a T-type converter used in PV
quencies (Elkayam & Kuperman, 2019). Even though, the applications is shown in Fig. 1, which consists of IGBT
use of a second-order generalized integrator (SOGI) (Song switches and a boost reactor. With the pulse width modu-
et al., 2016) as a coordinate transformation may cause a neg- lation (PWM) technique, the converter can be treated as an
ative effect on the dynamic of the control system. Sliding ideal voltage source. This means the system can be described
mode control (SMC) with excellent tracking performance, by:
appears to be a promising candidate to solve the problems of
the grid current control loop. With superior fast dynamic, the di L (t)
L = u s (t) − R L i L (t) − mUdc (1)
SMC can be designed to track the alternating (AC) current dt
directly (Linh & Phuong, 2021; Binh et al., 2020). Hence, dVdc (t)
a coordinate transformation can be neglected. Nevertheless, C = mi L (t) − i o (t) (2)
dt
practical issues such as modeling error, parameter variation,
and especially the chattering effect of SMC itself must be in which, u s (t), i L (t) and i o (t) are line voltage, inductor
taken into account in control design (Hao et al., 2013; Komur- current, and DC load current, respectively; −1 ≤ m ≤ 1 is
cugil et al., 2016). For single-phase grid-tie converters, the the modulation index; C = CC11+C C2
2
, and L are the DC-link
THD of the output current is strongly affected by the second capacitor and output inductor.
harmonic ripple across the DC bus. Because the reference A discrete-time version of (1) and (2) is derived by apply-
of the grid-tie current is computed by the DC bus voltage ing Forward-Euler with sampling time Ts , and results in:
regulator, properly designed for this outer-loop control also  
plays a very important role (Li et al., 2017). Otherwise, the RL Ts Ts
i L,k+1 = 1− Ts i L,k + u s,k − Vdc,k m k (3)
generated current may be polluted by this second-order har- L L L
monic. Besides, inrush current limit and voltage oscillation
mTs Ts
are also problems needed to be solved to extend the life- Vdc,k+1 = Vdc,k + i L,k − i o,k (4)
C C
time of DC-link capacitors. Another issue of the neural point
clamp (NPC)-based topologies is the unbalance problems Based on (3) and (4), a fully digital controller is designed
between the dc link capacitors. This issue has been consid- to satisfy the strict requirements of a grid-connected con-
ered in another research of us (Linh & Phuong, 2021), and verter.
hence is not going to be discussed in this paper in detail.

123
Journal of Control, Automation and Electrical Systems

Fig. 1 Configuration of a T-type converter in PV applications

3 Control Design tuation of the DC-bus voltage is minimized. Theoretically,

a conventional discrete-time SMC can solve the aforemen-
Considering the following voltage equation of the grid- tioned problem. The control design is as follows.
connected converter: First, define the sliding variable Sk as a function of the
inductor current tracking error ek
di L (t)
vg (t) = Ri L (t) + L + vinv (t) (5)
dt Sk = ek − λek−1 (6)

in which, vg (t) = Vg cos(ωt) is the grid voltage, vinv (t) is where ek can be calculated by:
the voltage generated by the inverter. The phasor diagram of
the system as cos ϕ = ±1 is shown in Fig. 2. ∗
ek = i Lk − i Lk (7)
As cos ϕ = ±1, the current generated by the inverter
always aligns with the grid voltage, i.e., i L (t) = I Ld cos(ωt) Then, by solving the following equation, the equivalent con-
with cos ϕ = 1 and i L (t) = I Ld cos(ωt + π ) with cos ϕ = trol signal which maintains the Sk on the sliding manifold
−1. Thus, by using a phase-lock-loop (PLL) to get the infor- can be found.
mation from the grid side, i.e., cos(ωt), then the reference
for the current controller is in fact the amplitude I Ld of the Sk+1 = 0 (8)
grid-connected current. Consequently, the block diagram of
the control system is as in Fig. 3, which consists of a PLL to By considering (7) in one step-ahead and substituting it into
synchronize with the grid, a DC-bus voltage regulator, and an (8), it gives
inner current controller. In addition, a notch filter is used to
∗
extract the average dc voltage (Vdc Filter) from the double- i L,k+1 − i L,k+1 − λek = 0 (9)
frequency oscillation hidden in its measurement (Vdc Fb).
The design procedure of each controller is going to be clari- Now, substitute (3) into (9), a simple manipulation yields
fied in the next subsections.
   
L ∗ R
m k,eq = − i L,k+1 − 1 − Ts i L,k
3.1 Inductor Current Controller Ts Vdc,k L
  (10)
L Ts
It can be said that the inductor current control loop plays a − − u s,k − λek
Ts Vdc,k L
key role in improving the performance of the grid-connected
converters. With a high accuracy and quick response current where −1 ≤ m k,eq ≤ 1 is an equivalent control signal.
controller, the power transfer between the renewable energy Practically, (10) cannot eliminate the tracking error due to
source and the grid can be achieved quickly and smoothly. the inevitable modeling error. Hence, an additional switching
Which results in a low THD inductor current whilst the fluc- control action is often used. Following that is the existence

123
 Journal of Control, Automation and Electrical Systems

Fig. 2 Phasor diagram of the

grid-connected converter as
cos ϕ = ±1

Fig. 3 The block-diagram of the grid-connected control system

of the chattering phenomenon. Besides, the amplitude of the The right-hand side of (11) can be rewritten in details as
switching control signal must be designed based on the upper  
bound of the perturbation which is normally unknown, also. ∗ ∗ RL
i L,k+1 − λi L,k = − Ts − 1 + λ i L,k
Hence, adaptive sliding mode control (ASMC) which can L
(12)
effectively deal with the parameter variation is employed in Ts  
+ u s,k − Vdc,k m k
this situation. To design the ASMC, rewrite (9) as follows L
In order to apply the adaptive control design, (12) is trans-
∗ ∗
i L,k+1 − λi L,k = i L,k+1 − λi L,k (11) ferred into vector form as following

123
Journal of Control, Automation and Electrical Systems

∗ ∗ 
i L,k+1 − λi L,k = kT θ0 (13) N
k−1
T 
k−1
2
k
(b) lim 2
<∞ (24)
N →∞ c + k−1
T 
k=1 k−1
in which,

N
 T (c) lim ||θ̂k − θ̂k−1 ||2 < ∞ (25)
θ0 = RL
Ts − 1 + λ Ts
(14) N →∞
L L k=1
 
kT = −i L,k u s,k − Vdc,k m k (15) 
N
(d) lim ||θ̂k − θ̂k−i ||2 < ∞ (26)
N →∞
k=n
If we could choose a control signal m k such that (13) is ful-
filled, it is obvious that the tracking error would be identically (e) lim ||θ̂k − θ̂k−i || = 0 (27)
k→∞
zero. In this situation, since θ0 is unknown, an estimation of
θ0 , i.e., θ̂k generated by an adaptive algorithm is used instead. where i is a finite positive integer.
Consequently, an adaptive control law can be obtained by
solving the following equivalence equation 3.3 Stability Analysis of the Current Control-Loop

∗ ∗
i L,k+1 − λi L,k = kT θ̂k (16) By considering (6), (13) and (17), a simple manipulation
gives:
Define k as:
Sk =k−1
T
θ̂k−1 − k−1
T
θ0
(28)
k = i L,k − λi L,k−1 − k−1
T
θ̂k−1 (17) =− k

Then by using the projection algorithm (Goodwin & Sin, Substituting (28) into (23) and squaring both sides, it yields
1984), it results in
Sk2
lim =0 (29)
αk−1 k→∞ c + k−1 k−1
T
θ̂k = θ̂k−1 + k (18)
c + k−1 Pk−1 k−1
T
To guarantee that the system is stable and the tracking error
with θ̂0 given; 0 < α ≤ 2, c > 0 are two constants, and approaches zero asymptotically, we must prove that ||k−1 ||
is bounded.
 T Suppose that the original SMC can stabilize the control
θ̂k = θ̂1,k θ̂2,k (19) system by (10), which means there always exists constants
0 < ai < ∞; (i = 1, 2) such that the following relation
Based on (16), (18), and (19), the adaptive control law between the modulation index m and the output current i L :
applied to the power converter is
 m k−1 ≤ a1 + a2 max|i L,τ − λi L,τ −1 |
1 ∗ ∗ (30)
mk = −i L,k θ̂1,k + θ̂2,k u s,k − i L,k+1 + λi L,k 1 ≤ k ≤ N; 1 ≤ τ ≤ N
Vdc,k θ̂2,k
(20) Based on (30), ||k−1 || can be rewritten as

3.2 Properties of the Adaptive Control k−1

T
k−1 = i L,k
2
+ (u s,k − Vdc,k m k )2
 (31)
According to Goodwin and Sin (1984), the update law (18) < 2 i L,k
2
+ u 2s,k + Vdc,k
2
m 2k
results in the following properties
Define A = k−1
T 
k−1 and substitute (30) into (31), it
(i)||θ̂k − θ0 || ≤ ||θ̂k−1 − θ0 || ≤ ||θ̂0 − θ0 || (21)
yields

N 2
(ii) lim k
<∞ (22) 
2
N →∞
k=1
c + k−1
T 
k−1 A < 2 u 2s,k + i L,k
2 2
+ 2Vdc,k a1 + a2 max|i L,k − λi L,τ −1 |

2
A < 2 u 2s,k + i L,k
2 2
+ 4Vdc,k a12 + a22 max|i L,τ − λi L,τ −1 |
which implies
(32)

2
(a) lim
k
=0 (23) A < 4 u 2s,k + i L,k
2 + V2 2 2
dc,k a1 + a2 max|i L,τ − λi L,τ −1 |
1
k→∞
c + k−1
T 
k−1
2
(33)

123
 Journal of Control, Automation and Electrical Systems

A < 4 u 2s,k + Vdc,k
2 a2
1 3.4 DC Bus Voltage Control

2 2 2 2
+ 4 i L,k + Vdc,k a2 max|i L,τ − λi L,τ −1 | (34) Suppose that there are no losses in the AC inductances and
the converter, the variation of the magnetic energy is ignored,
It should be noted that as well as the proposed ASMC is perfect to guarantee that
cos ϕ = ±1, the variation of the DC side capacitor energy
|i L,τ − λi L,τ −1 | < |i L,τ | + λ|i L,τ −1 | can be expressed by
→max|i L,τ − λi L,τ −1 | < 2max|i L,τ |
2 (t)
2 2 1 dVdc
→ max|i L,τ − λi L,τ −1 | < 4 max|i L,τ | (35) C = Pac (t) − Pdc (t) (41)
2 dt
Then, it can be derived from (34) and (35) that Define Req as the equivalent load of the DC bus, then the AC
  power Pac and the DC power Pdc can be computed by
2
A < 4 u 2s,k + Vdc,k
2
a12 + 4 1 + 4Vdc,k
2
a22 max|i L,τ |
1
(36) Pac (t) = Vg I Ld (t)
2
(42)
V 2 (t)
Because the following conditions always hold in practice: Pdc = dc
Req
Vdc,k > 0, u s,k > 0
(37) 2 (t) by V (t),
Now, based on (41), (42), and substitute Vdc
0 < a1 < ∞, 0 < a2 < ∞
it yields
Then, the inequality (36) can be rewritten as 1 d V (t) 1 1
C =− V (t) + Vg I Ld (t) (43)
   2 dt Req 2
√
A < 2 u s,k + Vdc,k a1 + 1 + 4Vdc,k
2 a2
2 max|i L,τ |
By applying Forward-Euler to (43), the discrete-time transfer
(38) function between the input I Ld,k and the output Vk can easily
be found as follows
Now, rewriting (6) in detail, it results in
Vg T s −1
Vk C z
∗ ∗ = (44)
− Sk = i L,k − λi L,k−1 − i L,k − λi L,k−1 I Ld,k 1− 1 − R2T s
z −1
eq C
∗ ∗
→ |Sk | ≥ |i L,k − λi L,k−1 | − |i L,k − λi L,k−1 |
∗ ∗
→ |i L,k − λi L,k−1 | ≤ |Sk | + |i L,k − λi L,k−1 | (39)
Theoretically, a simple proportional-integral (PI) controller
can be designed to stabilize the DC bus based on (44). The
→ max|i L,k − λi L,k−1 | ≤ 2max|i L,τ | ≤ |Sk | + a3
transfer function of the PI controller in the discrete-time
1 domain with sampling time Ts , proportional coefficient K P
→ max|i L,τ | ≤ (|Sk | + a3 )
2 and integral time constant TI is
with 0 < a3 < ∞.
1 − Dz −1
Substitute (39) into (38), it gives WP I = K P (45)
1 − z −1

√   with,
A < 2 u s,k + Vdc,k a1 + 1 + 4Vdc,k
2 a 2 max (|S | + a )
2 k 3

(40) Ts
D =1− (46)
TI
Based on (40) and the key technical lemma√ 6.2.1. in Goodwin
and Sin (1984), it can be concluded that A = ||k−1 || is The closed-loop transfer function of the DC bus voltage loop
bounded. Then Sk → 0, following that the tracking error with PI controller is
ek approaches zero asymptotically. It also follows from (22)
that the square of the tracking error is summable, and (27) b1 K P z −1 − b1 K P Dz −2
WC L =
shows that the estimated parameters are slow time-varying 1 + (b1 K P − a1 − 1) z −1 + (a1 − b1 K P D) z −2
as k → ∞. (47)

123
Journal of Control, Automation and Electrical Systems

V T
where b1 = gC s and a1 = 1 − R2T s
eq C
are parameters of PI controller is activated. This proposed controller eliminates
(44). the tracking error at the end of the second phase instantly with
Once the closed-loop transfer function is established, the negligible overshoot, as well as minimizes the dc-bus voltage
parameters of the controller can be obtained by various oscillation caused by the variation of the equivalent load Req .
model-based control techniques, such as pole-placement, and The tuning rule applied to the PI controller is inspired by the
optimization. However, it should be noted that the equivalent pole-placement method as follows.
load Req is unknown in practice, although its bound can be Assume that Req is known, the integral time constant Ti
obtained based on the rated power Prated of the converter and should be chosen such that the zero of PI controller (45) and
the loss power P0 = η Prated , where η is the efficiency of the the pole of the control object (44) cancel each other. Which
converter. means

2 2 Req C
Vdc,rated Vdc,rated Ti = (52)
≤ Req ≤ (48) 2
Prated P0
Consequently, the closed-loop transfer function of the DC
The problem caused by the unknown Req can be solved voltage control system is order-reduced as follows
as follows. Practically, the operation of the converter can be
classified into three phases. Suppose that the system is in an Ts K p Vg −1
C z
idle state, which means all the dc-bus capacitors are empty. To WC L = Ts K p Vg −1
(53)
(1 − −1
z )+ C z
startup, the DC bus capacitor C is charged via a serial resistor
and the diode rectifier in the first phase. The inrush current
In this ideal case, the dynamic of the voltage loop is mainly
in this phase is restricted by an appropriate resistor, and this
decided by K p , i.e., the larger the K p , the faster the transient
simple approach is widely used in commercial devices.
response. However, Req changes with the transferred power
As the dc voltage reaches its natural stable value Vg , the
in practice. Therefore, it is impossible to achieve pole-zero
serial resistor is bypassed, and the system moves to the second
cancellation. On the other hand, choosing fixed Ti may sta-
phase where the controllers are activated to boost the voltage
bilize the control system, but the performance may not be
to its desired value. During this phase, the load is not allowed
satisfactory. For example, small Ti in the light-load condi-
to connect to the dc-bus, following that (44) can be regarded
tion may cause a large overshoot in the transient state, or
as a pure integrator. Using the PI controller in this phase may
large Ti yields a slow dynamic response to a sudden change
cause a large overshoot and fluctuation in the DC bus due to
of power transferred. Hence, a fuzzy-based PI controller is
the existence of the integral. Hence, a proportional controller
proposed in this case to adjust both parameters K p and Ti
is sufficient in this phase. The inrush current amplitude can
according to working conditions.
be limited by choosing the proportional gain K p as follows.
Inspiring by the pole-placement method, it can be seen
Assume that E v,max is the maximum error when the volt-
from (52) and (48) that the integral time constant should also
age control-loop is activated:
be bounded by
E v,max = V ∗ − Vm2 (49) 2 2
C Vdc,rated C Vdc,rated
≤ Ti ≤ (54)
Obviously, the inrush current is maximum at this instance. 2Prated 2P0
Since a proportional controller is employed, the maximum In steady-state, the main goal is to eliminate the tracking
∗
reference value I Ld,max can easily be computed by error. Hence, a small gain K P and large integral time Ti is
sufficient. In the transient state where a sudden change of
∗
I Ld,max = K p E v,max (50) transferred power occurs, large gain K p and small Ti should
be used to quickly suppress the fluctuation of the dc voltage.
As a results, the current used to boost the capacitor voltage The aforementioned discussion, in combination with restric-
is limited by selecting K p as tions (48) and (54), is formularized by Mamdani fuzzy rule
∗
with triangle membership function (MF) as follows
I Ld,max
Kp = (51)
E v,max rule 1: If |ek | is PZ Then K p is PZ and Ti is PH
rule 2: If |ek | is PS Then K p is PS and Ti is PL
With the proposed proportional controller, the dc-bus volt-
age reaches its set point asymptotically. And at the end of the rule 3: If |ek | is PM Then K p is PM and Ti is PM (55)
second phase when the tracking error is sufficiently small, the rule 4: If |ek | is PL Then K p is PL and Ti is PS
control system moves to the third phase where a fuzzy-based rule 5: If |ek | is PH Then K p is PH and Ti is PZ

123
 Journal of Control, Automation and Electrical Systems

Fig. 4 Fuzzy controller design

Table 1 Parameters of the grid-connected converter

⎧ 2x−2c +w w Element Value Element Value Element Value
r ,x r ,x
⎪
⎨ wr ,x f or cr ,x − 2r ,x ≤ x ≤ cr ,x
M Fr (x) = 2cr ,x −2x+wr ,x f or cr ,x < x ≤ cr ,x + wr ,x Lf 0.84 mH C1 , C2 940μF Prated 3 kW
⎪
⎩ wr ,x 2
Rf 0.05 Ts 100 μs ∗
Vdc 400 V
0 otherwise
Rc 0.05 f sw 10 kHz η 0.95
(56)
α 0.3 λ 0.9 c 1

where r = {1, · · · , 5} is the index of the rule, x =

{|ek |, K p , Ti } stands for the input variables of the MFs, cr ,x
and wr ,x are center and width of each MF, and the fuzzy set
is designed as shown in Fig. 4 such that consistency between simulation and future practical
Finally, by using the center of gravity method for defuzzi- implementation is assured. Parameters used in the simulation
fication, and noting that all MFs are symmetrical triangles, it are provided in Table 1.
yields The starting sequence of the converter is illustrated in
Fig. 5, exactly the same as described in section II.C. As can

5
be observed, with a well-designed notch filter, the average
M Fr (|ek |)w(r ,Ti,k ) c(r ,Ti,k )
r =1 dc voltage (Vdc filter) is exactly extracted from its instan-
Ti,k = (57)

5 taneous (Vdc Fb), and quickly tracks its desired one (Vdc
M Fr (|ek |)w(r ,Ti,k )
r =1
Ref) with minimized overshoot. The gain K p is chosen such

5 that the amplitude of the charging current in Phase − 2 is
M Fr (|ek |)w(r ,K p,k ) c(r ,K p,k ) restricted at 15A. At time instance t = 0.2s, when the fuzzy-
r =1
K p,k = (58) PI controller has already been activated, results shown in

5
Fig. 5 with sudden power jump, i.e., P = 0 to P = 3kW and
M Fr (|ek |)w(r ,K p,k )
r =1 P = 3kW to P = −3kW , shows that the DC bus voltage
fluctuation is quickly suppressed in less than 20ms. In addi-
tion, the overshoot is also restricted below 1%. This response
4 Numerical Simulations is much faster than the results achieved in Li et al. (2017).
The power transferred and the variation of K p and Ti in each
To verify the effectiveness of the proposed solution, vari- phase are shown in Figs. 6 and 7, respectively.
ous numerical simulations are conducted in this section, and The performance of the ASMC is shown in Fig. 8. It can be
all controllers are implemented using standard C language seen that the average value of the measured current tracks its

123
Journal of Control, Automation and Electrical Systems

Fig. 5 Startup sequence of the

grid-connected converter

Fig. 6 Power transferred

between the renewable energy
source and the grid line

Fig. 7 Variation of K p and Ti

according to fuzzy tuning

123
 Journal of Control, Automation and Electrical Systems

Fig. 8 Tracking performance of

the DASMC

Fig. 9 Parameters convergence

of the DASMC

Fig. 10 Robustness of the

proposed control system against
parameters variation and
harmonic

123
Journal of Control, Automation and Electrical Systems

desired one instantly. Meanwhile, all the parameters quickly Table 2 Parameters of the LCL filter
converge as shown in Fig. 9. Element Value Element Value
The most important feature of any controller is robustness
against parameter variation, which shows its ability to be Vac 220 V r ms f res 5000 Hz
used in practice, Particularly for grid-connected converters, Li 1.5 mH Lg 0.11 mH
the inductance and DC bus capacitance are always decreased Cf 10 μF
after a long time of operation. The tolerance may be up
to 10% → 20%. Hence, simulation in the worst situation,
where both sudden changes of parameters and harmonics of LCL filter is designed (Reznik et al., 2014) and the results
line voltage occur simultaneously, is carried out to verify are provided in Table 2.
the robustness of the proposed control strategy. As shown in which, C f is the filter capacitor, L i and L g are inverter
in Fig. 10, the stability of the control system is guaranteed, side and grid side inductor, respectively.
although i L is distorted due to the inductance attenuation. The simulation results with output LCL filter are shown in
Since most of the output filter. With input constraints such Figs. 11 and 12, with initial inductance for the ASMC is half
that: current ripple i is below 5% of the rated current, reac- of the sum of L i + L g . It can be observed from Fig. 12 that
tive power exchange when the inverter is not operating is 5% with the designed LCL filter, the THD of the grid-connected
of the rated power, and the resonance frequency of the grid- current is improved, i.e., 3.41% in comparison with 5.25%.
side filter f res satisfies 10 f sw < f res < f2res , the standard The response of both the voltage and current controllers is

Fig. 11 Simulation results with

Injected current(A)
400 60
Line voltage(V)

Line voltage
output LCL filter 200 Injected current 30

0 0

-200 -30

-400 -60
0 0.1 0.2 0.3 0.4 0.5 0.6
Time(s)
Dc bus voltage(V)

400 Vdc Ref

Vdc Fb
Vdc Filter
200

0
0 0.1 0.2 0.3 0.4 0.5 0.6
Time(s)

Fig. 12 THD analysis of L and LCL filters

123
 Journal of Control, Automation and Electrical Systems

Fig. 13 Capacitor voltage 250

balance Vc1
Vc2
200

Capacitor voltage(V)
205
150
200
195
100
0.424 0.426 0.428

50

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Time(s)

similar to which getting from the configuration with the sim- Acknowledgements This research is funded by the Hanoi University of
ple L filter. This property can be explained based on the fact Science and Technology (HUST) under project number T2022-PC-005.
that a well-designed L g C f forms a low-pass filter that fil-
ters out the switching frequency. Thus, the impedance of the Declarations
filter capacitor at grid frequency is large, following that the
influences of the current flowing through the capacitor can Conflict of interest The author declares no conflict of interest.
be neglected. Whilst the parameters of the ASMC are con-
tinuously adjusted to minimize the tracking error. Hence, the
proposed ASMC can operate with both simple L and LC L
filters without any changes in the control design. Finally, References
Fig. 13 shows that with the PWM strategy proposed in Linh
and Phuong (2021), the voltage of DC-bus capacitors is kept Elkayam, M., & Kuperman, A. (2019). Optimized design of mul-
tiresonant AC current regulators for single-phase grid-connected
balanced with a small ripple although the double grid fre- photovoltaic inverters. IEEE Journal of Photovoltaics, 9(6), 1815–
quency oscillation is inevitable in this case. 1818. https://doi.org/10.1109/JPHOTOV.2019.2937386
Goodwin, G. C., & Sin, K. S. (1984). Adaptive filtering prediction and
control. Prentice-Hall.
Guo, X., & Guerrero, J. M. (2016). General unified integral controller
with zero steady-state error for single-phase grid-connected invert-
ers. IEEE Transactions on Smart Grid, 7(1), 74–83. https://doi.
5 Conclusions org/10.1109/TSG.2015.2419236
Hao, X., Yang, X., Liu, T., Huang, L., & Chen, W. (2013). A sliding-
mode controller with multi resonant sliding surface for single-
In this study, a solution to improve the performance of single- phase grid-connected VSI with an LCL filter. IEEE Transactions
phase grid-connected converters is discussed. For the inner on Power Electronics, 28(5), 2259–2268. https://doi.org/10.1109/
current control loop, a discrete-time ASMC is employed to TPEL.2012.2218133
achieve multiple goals such as quick response, good tracking Kjaer, S. B., Pedersen, J. K., & Blaabjerg, F. (2005). A review of single-
phase grid-connected inverters for photovoltaic modules. IEEE
performance, and robustness against parameter variation. For Transactions on Industry Applications, 41(5), 1292–1306. https://
the outer dc-bus voltage control loop, limiting the inrush cur- doi.org/10.1109/TIA.2005.853371
rent and voltage overshoot play an extremely important role Komurcugil, H., Ozdemir, S., Sefa, I., Altin, N., & Kukrer, O.
in the lifetime of the capacitors. With the proposed control (2016). Sliding-mode control for single-phase grid-connected
LCL-filtered VSI with double-band hysteresis schemel. IEEE
strategy, in which the configuration, as well as the parameters Transactions on Industrial Electronics, 63(2), 864–873. https://
of the controller, are tuned by fuzzy law, both aforementioned doi.org/10.1109/TIE.2015.2477486
goals can be achieved. Results getting from numerical simu- Li, B., Huang, S., & Chen, X. (2017). Performance improvement for
lations show that the overshoot of the dc bus voltage no longer two-stage single-phase grid-connected converters using a fast DC
bus control scheme and a novel synchronous frame current con-
exists in startup mode and is less than 1% in the transient state. troller. Energies, 10(3), 389. https://doi.org/10.3390/en10030389
In addition, the stability of the system is guaranteed with both Linh, N., & Phuong, V. (2021). Advanced single-loop discrete-time
simple L and standard LC L filter, and even with line voltage control for T-type voltage source inverterwith minimum capacitor
polluted by harmonic, as well as ±20% of parameters toler- voltage ripple modulation. Turkish Journal of Electrical Engineer-
ing and Computer Sciences. https://doi.org/10.3906/elk-2010-101
ances. More experiments with prototype hardware are going Pho, B. B., Tran, T. M., Nguyen, M. L., & Hoang, P. V. (2020). Discrete-
to be carried out in near future. Time Quasi Sliding Mode Control of Single-phase T-type Inverters
for Residential PV Applications. In International Conference

123
Journal of Control, Automation and Electrical Systems

on Advanced Mechatronic Systems (ICAMechS) (pp. 118–123). Tsili, M., & Papathanassiou, S. (2009). A review of grid code technical
https://doi.org/10.1109/ICAMechS49982.2020.9310085 requirements for wind farms. IET Renewable Power Generation,
Reznik, A., Simões, M. G., Al-Durra, A., & Muyeen, S. M. (2014). LC L 3(3), 308–332. https://doi.org/10.1049/iet-rpg.2008.0070
filter design and performance analysis for grid-interconnected Wang, H., Cheng, Q., Li, M., Chen, G., & Deng, L. (2014). The study
systems. IEEE Transactions on Industry Applications, 50(2), of single-phase PWM rectifier based on PR control strategy. In
1225–1232. https://doi.org/10.1109/TIA.2013.2274612 The 26th Chinese Control and Decision Conference (CCDC) (pp.
Silwal, S., Taghizadeh, S., Karimi-Ghartemani, M., Hossain, M. J., & 3818–3823). https://doi.org/10.1109/CCDC.2014.6852845
Davari, M. (2019). An enhanced control system for single-phase Zeb, K., Khan, I., Uddin, W., Khan, M. A., Sathishkumar, P., Busarello,
inverters interfaced with weak and distorted grids. IEEE Transac- T. D. C., Ahmad, I., & Kim, H. J. (2018). A review on recent
tions on Power Electronics, 34(12), 12538–12551. https://doi.org/ advances and future trends of transformerless inverter structures
10.1109/TPEL.2019.2909532 for single-phase grid-connected photovoltaic systems. Energies,
Song, W., Deng, Z., Wang, S., & Feng, X. (2016). A simple model 11(8), 1968. https://doi.org/10.3390/en11081968
predictive power control strategy for single-phase PWM convert-
ers with modulation function optimization. IEEE Transactions
on Power Electronics, 31(7), 5279–5289. https://doi.org/10.1109/ Publisher’s Note Springer Nature remains neutral with regard to juris-
TPEL.2015.2481323 dictional claims in published maps and institutional affiliations.
Sulaeman, I., Vega-Garita, V., Mouli, G. R. C., Narayan, N., Ramirez-
Elizondo, L., & Bauer, P. (2016). Comparison of PV-battery Springer Nature or its licensor (e.g. a society or other partner) holds
architectures for residential applications. In IEEE International exclusive rights to this article under a publishing agreement with the
Energy Conference (ENERGYCON) (pp. 1–7). https://doi.org/10. author(s) or other rightsholder(s); author self-archiving of the accepted
1109/ENERGYCON.2016.7514014 manuscript version of this article is solely governed by the terms of such
publishing agreement and applicable law.

123