energies

Article
Voltage-Sensorless Control Scheme for a Grid
Connected Inverter Using Disturbance Observer
Hyun-Sou Kim and Kyeong-Hwa Kim *
Department of Electrical and Information Engineering, Seoul National University of Science and Technology,
232 Gongneung-ro, Nowon-gu, Seoul 01811, Korea; apd142@hanmail.net
* Correspondence: k2h1@seoultech.ac.kr; Tel.: +82-2-970-6406; Fax: +82-2-978-2754

Academic Editor: Adolf Acquaye

Received: 1 November 2016; Accepted: 23 January 2017; Published: 2 February 2017

Abstract: A grid connected inverter usually requires voltage and current measurements to control
the active and reactive powers as well as the inverter output currents. While voltage sensors are
essential to obtain reliable information on the phase angle, these additional components certainly
increase the production costs and complexity. In this paper, a voltage-sensorless control scheme for
a grid connected inverter using a disturbance observer (DOB) is presented. The grid voltages are
estimated by DOB in the stationary reference frame using the current measurements and reference
signals. Even though the DOB estimates the grid voltages with reasonable accuracy in the presence
of the uncertainty such as the unbalanced condition and harmonic distortion, the resultant waveform
shows a phase lag depending on the estimation bandwidth. To overcome this limitation, a phase
lead compensation is introduced. By using these techniques, the phase angle of grid voltages
can be completely restored even if the phase angle of grid is initially unknown. The proposed
scheme is simple and straightforward. In addition, it does not require any additional hardware.
The feasibility of the proposed voltage-sensorless control scheme is demonstrated through simulations
and experiments using 2 kVA prototype inverter.

Keywords: distributed generation (DG); disturbance observer (DOB); estimation of grid voltages
and phase angle; grid connected inverter; voltage-sensorless control

1. Introduction
Recently, distributed generations (DGs) using renewable energy resources, which include the
wind power and photovoltaic generation, have been received an increased attention. A primary
objective in future electrical network is to maximize the penetration of DG systems in constructing the
microgrid in a way that enhances the overall grid stability and reliability [1]. Because most DGs operate
in a grid connected mode, single-phase or three-phase grid connected inverters have become key
power electronic interfaces between DG and the utility grid [2–10]. The main goal of grid connected
inverters is to inject the current into the electrical power grid with high performance in the sense of a fast
dynamic response, robustness to disturbance, and zero tracking error. The power quality of the injected
current should be sufficiently high even if the grid voltages encounter uncertain disturbances such as
the harmonic distortion and imbalance. In addition to those properties, the pulsewidth modulation
(PWM) grid connected inverter should provide essential functions such as the grid synchronization,
the control of the active power, and the injection of the reactive power under specific condition to meet
the grid codes.
Undesirable harmonic contents in electrical system disrupt power transformer operation and
produce excessive power dissipation in the transmission lines [2]. Moreover, the harmonic pollution
leads to many critical problems such as the distorted grid voltages, malfunction of the high accuracy
equipment, and damage on the electrical systems [11]. For this reason, the harmonic restriction

Energies 2017, 10, 166; doi:10.3390/en10020166 www.mdpi.com/journal/energies

Energies 2017, 10, 166 2 of 19

standards such as IEEE-519 or IEC 61000-3-2 have been published to limit the amount of harmonic
current injected into utility grid [12]. In the same vein, many researchers have investigated the
strategies to improve the power quality of the grid connected inverter by mitigating the current
harmonics [11–18]. As a selective harmonic compensation scheme, a proportional integral (PI)-resonant
current controller has been presented to compensate low-order harmonic terms selectively [13,16].
As an alternative method, nonselective harmonic compensation techniques such as the sliding mode
control [11], predictive control [17], and repetitive control [18] have been designed.
As the expansion of DG systems progresses rapidly, the economic feasibility such as the
manufacturing cost and system size has become another desirable feature in grid connected inverters.
Less feedback sensors often lead to smaller size of hardware and lower cost [1]. To maximize the
efficiency and performance, most of grid connected inverters employ three types of sensors in their
structure which are the DC link voltage sensor, the grid voltage sensors, and current sensors [19].
Meanwhile, there have been attempts to eliminate some of these sensors in power electronics
converters [19–23]. These studies pursue to reduce the system complexity in PWM rectifiers or inverters,
while still retaining the system performance and reliability. To allow the grid voltage sensorless
operation in single-phase grid connected converters, the voltage estimation scheme based on the
measured currents and converter parameters has been proposed [20]. The voltage estimation from
the estimated instantaneous active and reactive powers has been studied for PWM converters [21].
However, since this scheme requires numerical differentiation, high-frequency noises may degrade the
estimation accuracy.
A nonlinear Luenberger type observer has been presented to eliminate the line voltage sensors in
three-phase voltage source converter [22]. To estimate the equivalent grid impedance and equivalent
grid voltage, an extended Kalman filter has been introduced [23]. Although these schemes give
a relatively good estimation performance, it is not easy to implement these schemes in commercial
systems due to the algorithm complexity and computational burden. Furthermore, these methods
have not been verified under various grid conditions including the grid uncertainties such as the
imbalance and harmonic distortion.
As another approach, numerical algorithms such as the Newton–Raphson optimization, was
introduced to estimate the grid voltage in three-phase grid connected voltage source converters [19].
However, it is well known that this scheme is quite complicated because nonlinear interconnected
equations should be solved by iteration. In addition, the convergence should be carefully considered
to get the solution with an acceptable accuracy. Grid synchronization method without the line voltage
sensors has been presented by using an adaptive observer to estimate unknown dynamics [1].
As an effective way of observing external disturbances in systems, the estimation principle based on
DOB has recently been studied [24]. In theory, the DOB uses the information on the output measurement
and known control input together with the system inverse model to estimate the unknown disturbances.
In view of the state equation model of a grid connected inverter, the grid voltages can be regarded as
the disturbance, which enables the estimation of grid voltages through the DOB.
To retrench the manufacturing cost without the loss of reliability, this paper proposes
a voltage-sensorless control scheme for a grid connected inverter using DOB. The proposed scheme
mainly consists of two parts: a current control design based on the proportional-resonant (PR)
controller and the estimator design of grid voltages based on the DOB. Generally, to synchronize
the grid connected inverters to utility grid, a phase-locked loop (PLL) algorithm using the measured
voltages is applied because of its simplicity and performance. To obtain the grid voltage information
without voltage sensors, the proposed scheme employs the DOB based estimation using the current
measurements and reference signals. The estimated grid voltages exhibit the phase lag from actual
grid voltages due to the bandwidth limit of the DOB. This phase delay is compensated before the
estimated grid voltages are used in the PLL algorithm to determine the phase angle of grid voltages.
To validate the effectiveness of the proposed voltage-sensorless control scheme, 2 kVA prototype
grid connected inverter has been constructed using the digital signal processor (DSP) TMS320F28335
Energies 2017, 10, 166 3 of 18

grid connected inverter has been constructed using the digital signal processor (DSP) TMS320F28335
(Texas Instruments, Dallas, TX, USA) [25]. Through the comparative simulations and experiments, it
Energies 2017, 10, 166 3 of 19
is verified that the proposed scheme works usefully with a reasonable reliability even under the grid
uncertainties such as the imbalance or harmonic distortion.
(Texas Instruments, Dallas, TX, USA) [25]. Through the comparative simulations and experiments, it is
2. Modeling
verified of aproposed
that the Grid Connected
schemeInverter
works usefully with a reasonable reliability even under the grid
uncertainties
Figure 1such as the
shows theimbalance or harmonic
configuration distortion.
of a grid connected inverter connected to the utility grid
through an L filter. In the stationary reference frame, the mathematical model of three-phase grid
2. Modeling of a Grid Connected Inverter
connected inverter is described as follows:
Figure 1 shows the configuration of a grid connected inverter connected to the utility grid through
vα = the
an L filter. In the stationary reference frame, + Liα + eα model of three-phase grid connected
Riαmathematical (1)

i + L. i + e
inverter is described as follows:
=R
vv β = Riαβ + Liαβ+ eαβ
(2)
(1)
α
.
= Riβ + Lreference
where the subscript “αβ” denotes thevβstationary iβ + eβ frame; vα and vβ are the q-axis and
(2)

where
d-axis the subscript
inverter “αβ”in
voltages denotes the stationary
the stationary reference
frame, frame;iαvαand
respectively; andiβvβare
arethe q-axis and
theq-axis and d-axis
inverter voltages in the stationary frame, respectively; iα and iβ are the q-axis and d-axis inverter
inverter currents in the stationary frame, respectively; e and eβ are the q-axis and d-axis grid
currents in the stationary frame, respectively; eα and eβ areαthe q-axis and d-axis grid voltages in the
voltages in the stationary frame, respectively; and R and L represent the equivalent
stationary frame, respectively; and R and L represent the equivalent resistance resistance
and inductance ofand
the
inductance
filter of the filter
and connection and connection cables.
cables.

Figure 1. Configuration of a grid connected inverter.

Figure 1. Configuration of a grid connected inverter.

To regulate the active and reactive powers of DG system, the current of a grid connected
To regulate
inverter should the
be active and reactive
controlled with thepowers of DG system,
synchronization to the
gridcurrent of aTo
voltage. grid connected
alleviate the inverter
system
should be controlled with the synchronization to grid voltage. To alleviate the
instability due to the absence of phase angle information at the initial stage, the proposed system instability
scheme
due
uses toa the
PR absence
current of phase angle
controller information
in the stationaryatreference
the initialframe,
stage, which
the proposed scheme
contributes to auses a PR
stability
current controller in the stationary reference frame, which contributes
enhancement by minimizing the transformation with phase angle information. to a stability enhancement by
minimizing the transformation with phase angle information.
In the stationary reference frame, a PR current controller is expressed as follows:
In the stationary reference frame, a PR current controller is expressed as follows:
*
ur (s) = C(s) (iαβ (s) − iαβ (s)) (3)
∗
ur (s) = C (s)(iαβ (s) − iαβ (s)) (3)
* * * T T
where i αβ = [iα iβ ] , i αβ = [iα iβ ] , the symbol “*” denotes the reference quantities, s is a
∗
where iαβ ∗
= [ iα ∗ ]T , i
iβ
T
αβ = [ iα iβ ] , the symbol “*” denotes the reference quantities, s is
Laplace operator, and ur is the output of the PR controller. In Equation (3), C (s) is the transfer
a Laplace operator, and u r is the output of the PR controller. In Equation (3), C ( s ) is the transfer
function of the PR controller expressed as:
function of the PR controller expressed as:

K ωs
C =K
C ((ss)) = KPP++ 2
K RRωc sc
2 (4)
(4)
s + 2ω s + ω
s + 2ωc sc + ω2o o
2

where KKP Pand R are the proportional and resonant gains of the PR controller, respectively; ωc is
andK RKare
where the proportional and resonant gains of the PR controller, respectively; ωc is the
bandwidth; and ωo is the center frequency. To determine the eventual reference voltages of a grid
the bandwidth;
connected inverter,
ωogrid
andthe is the centerare
voltages frequency.
added into Toudetermine the eventual reference voltages of a
r by a feedforward manner as follows:
grid connected inverter, the grid voltages are added into ur by a feedforward manner as follows:
u∗ = ur + eαβ (5)
*
u = ur + eαβ (5)
T
where eαβ = [ eα eβ ] and u∗ = [ vα ∗ vβ∗ ] T is the reference voltages.

whereThee αβ
reference eβ ]T andare
= [eα voltages u* applied
= [vα* through
vβ* ]T is the
thereference
symmetrical space vector PWM technique.
voltages.
For further development of the DOB using the system inverse model, it is simply assumed that
the reference voltages are applied to the terminals of the inverter without any loss and deformation in
Energies 2017, 10, 166 4 of 18

The reference voltages are applied through the symmetrical space vector PWM technique. For
further development of the DOB using the system inverse model, it is simply assumed that the
Energies
reference 10, 166 are applied to the terminals of the inverter without any loss and deformation
2017,voltages 4 of in
19

the PWM inverter, that is, vα* = vα and vβ* = vβ . In addition, from Equations (1) and (2), the system
∗ ∗ = v . In addition, from Equations (1) and (2), the system
the PWM
model of ainverter,
grid connected α = vα is
that is, vinverter and vβ
rewritten βas follows:
model of a grid connected inverter is rewritten as follows:
i αβ = P ( s) (v αβ − e αβ ) (6)
iαβ = P (s) (vαβ − eαβ ) (6)
1 1 0
where v αβ = [vα vβ ]T , P(s) = I , and I = "  . #
Ls + R 0 11 0
] , P(s) = Ls1+ R I, and I =
T
where vαβ = [ vα vβ .
0 1
3. Estimation of Grid Voltage with Disturbance Observer
3. Estimation of Grid Voltage with Disturbance Observer
In this section, a DOB based control structure and grid voltage estimation process are described.
FromIn this section,
Equations a DOB
(1) and based
(2), the control
state structure
equation and
of a grid grid voltage
connected estimation
inverter process are
can be expressed as described.
follows:
From Equations (1) and (2), the state equation of a grid connected inverter can be expressed as follows:
y = A y + B u − Bd (7)
.
y = Ay + Bu − Bd (7)
R 1
where y = iαβ = [iα iβ ]T T, A = − I , B = I , and d =eαβ is regarded as an external
where y = iαβ = [ iα iβ ] , A = − RL 1 L
L I, B = L I, and d = eαβ is regarded as an external disturbance
disturbance
in in the state equation.
the state equation.
The objective of
The objective of the
the control
control design
design isis to
to steer
steer the
the state y to
state y to its
its desired
desired reference
reference value
value inin the
the
presence of external disturbancedd such
external disturbance such asasthetheharmonics
harmonicsand andimbalance,
imbalance,and andat at
thethe
samesame time,
time, to
to obtain
obtain thethe phase
phase angle
angle information
information by by estimating
estimating thethe
gridgrid voltages
voltages with with
the the
DOB.DOB. Figure
Figure 2 shows
2 shows the
the overall
overall blockblock diagram
diagram for the
for the proposed
proposed voltage-sensorless
voltage-sensorless control
control scheme.
scheme. TheThe
basicbasic concept
concept of
of the
DOB is that if the disturbed system input u − d can be estimated, the disturbance d can
the DOB is that if the disturbed system input u − d can be estimated, the disturbance d can also be
estimated by comparing uu−−ddwith withthe
theknown
knowncontrol
controlinput u [26].
inputu [26].

Figure
Figure 2.
2. Overall
Overall block
block diagram
diagram for
for the
the proposed
proposed voltage-sensorless control scheme.
voltage-sensorless control scheme.

From Equation (7), the disturbed control input can be calculated as follows:
From Equation (7), the disturbed control input can be calculated as follows:
u − d = B−−11( y. − A y) (8)
u − d = B (y − Ay) (8)
with u = v αβ in Equation (6), the disturbed control input u − d can be expressed as
with u = vαβ in Equation (6), the disturbed control input u − d can be expressed as
u − d = P−−11(s) y (9)
u−d = P (s)y (9)
Theoretically, u − d can be calculated from the output y if the system inverse model P ( s )
−1

Theoretically, u − d can be calculated from the output y if the system inverse model P−1 (s) is
−1
is realizable.
realizable. However,
However, because
because P−P1 ( s )(is
s )not
is not proper
proper in general,
in general, a stable
a stable Q(s)Qsuch
filterfilter (s) such as apass
as a low low
− 1
that Pthat(s)PQ−(1 s()s )isQproper.
filter (LPF)(LPF)
pass filter is required in theinimplementation
is required the implementation of theofDOB in order
the DOB in order ( s ) is proper.
Selecting Q as the first-order LPF, the disturbed system input u − d is now modified to the filtered
Selecting Q as the first-order LPF, the disturbed system input u − d is now modified to the
value as follows:
filtered value as follows:
u − d = P −1 ( s ) Q ( s ) y (10)
Energies 2017, 10, 166 5 of 19

Considering the first-order LPF and Equation (8), the state space expression for the output of the
transfer matrix P−1 (s) Q(s) is obtained as follows [26]:
.
q = Aq q + Bq y (11)
. .
û p = B−1 (y − Ay)= B−1 (q − A q) = B−1 ( Aq q + Bq y − A q) (12)

where Aq = − aτ0 I , Bq = aτ0 I , the symbol “ˆ” denotes the estimated quantities, q is the state of
the first-order LPF Q(s), û p is the output of the transfer matrix P−1 (s) Q(s), and a0 and τ are the
filter parameters, respectively. In Equation (12), the output û p corresponds to the filtered estimate
of u − d, which is determined by processing the output y with the transfer matrix P−1 (s) Q(s) as in
Equation (10).
Similarly, the same first-order LPF Q(s) is needed to process the known control input u. With the
same transfer matrix in Equation (10), the state space representation can be expressed as follows:
.
p = Aq p + Bq u (13)

u+ = p (14)

where p is the state of the LPF Q(s) and u+ denotes the output of the LPF. The output u+ represents
the filtered estimate of the known control input u. Since two filtered estimates for u − d and u were
obtained, the estimated disturbance that is the grid voltages in the stationary frame can be finally
calculated as:

dˆ = u+ − û p (15)

4. Proposed Voltage-Sensorless Control of a Grid Connected Inverter

In this section, the proposed voltage-sensorless control scheme which estimates the phase angle
of the grid voltage is explained using Figure 2. The q-axis and d-axis current references idq ∗ are

first transformed into the stationary values using the estimated phase angle. Using the processes
described in Section 2, the PR current controller is designed for a grid connected inverter in the
stationary reference frame. At the same time, the grid voltages are estimated through the DOB by
comparing the reference voltages with the current information processed with the system inverse
model. The estimated grid voltage êαβ in Equation (15), which is the output of the DOB, is used to
compose the reference voltage signal u as well as to generate the phase angle θ̂ through the synchronous
reference frame PLL (SRF-PLL).
However, due to the LPF Q(s) used in the implementation of the DOB, the estimated grid voltage
êαβ has some phase delay from the actual grid voltage eαβ . Such a phase delay is caused by the
limits of the DOB, which generally depends on the bandwidth of the LPF. Fortunately, the PR current
controller with the feedforward of the estimated grid voltage does not deteriorate the current control
quality severely even if the estimated grid voltage has small phase delay. Moreover, the frequency
variation is not significant in most grids. Thus, the phase angle determined from the estimated grid
voltage with the phase delay needs to be compensated.
From the LPF Q(s) in Equation (11), the transfer function of the first element can be obtained
as follows:

a0 /τ
Q11 (s) = (16)
s + a0 /τ
From Equation (16), the phase delay at the grid frequency introduced by the LPF can be simply
calculated as follows:
θdelay = ∠Q11 ( jωg ) = − tan−1 (τωg /a0 ) (17)
Energies 2017, 10, 166 6 of 19

where ωg denotes the grid angular frequency and θdelay denotes the phase delay due to the LPF in
DOB. Let θ̂ denote the phase angle determined from the estimated grid voltage êαβ . Because êαβ has
the phase delay θdelay from eαβ , the phase angle of the actual grid voltage eαβ can be simply obtained
by adding the delay angle θdelay as follows:

θ̂c = θ̂ + θdelay (18)

where θ̂c is the compensated phase angle which completely reconstructs the phase angle of the actual
grid voltage. This phase angle θ̂c is used to transform the current references into the stationary values
in Figure 2.
According to the harmonic limit on IEEE Standard 519-1992, the fifth- and the seventh-order
harmonics on the inverter output currents should be less than 4% [27]. To ensure the robustness against
harmonic distortions in the grid voltages, the harmonic regulators are incorporated in addition to
Equation (4). Then, the reference voltages of the proposed controller are obtained using the estimated
grid voltage êαβ as follows:

u∗ = urp + êαβ (19)

∗
urp (s) = C p (s)(iαβ (s) − iαβ (s)) (20)
K R n ωc s
C p (s) = K P + ∑ s2 + 2n ωc s + (n ωo )2
(21)
n=1,5,7

5. Simulations and Experimental Results

To prove the feasibility of the proposed voltage-sensorless control scheme for a grid connected
inverter, the simulations and experimental results are presented. The simulations are carried out using
the PSIM Software (Powersim, Rockville, MD, USA).
Figure 3 shows the configuration of the overall system. The entire system consists of a DC
source, a grid connected inverter, a DSP based controller, and three-phase programmable AC power
source that can emulate the ideal grid as well as distorted or unbalanced grid. The whole control
algorithm is implemented using DSP TMS320F28335 for 2 kVA prototype three-phase grid-connected
inverter [25]. The sampling period is chosen as 100 µs both in the simulations and experiments,
which yields the switching frequency of 10 kHz. The intelligent power module is employed for
three-phase grid-connected inverter. The inverter phase currents are detected by the Hall-effect devices
and are converted through internal 12-bit A/D converters, where the resolution of current is 18/211 A.
Figure 4 shows the experimental test setup consisting of a DSP based controller, three-phase grid
connected inverter, magnetic contactor (MC) for grid connection, and three-phase programmable AC
power source. The system parameters of the experimental system are listed in Table 1.
Energies 2017, 10, 166 7 of 18

Figure 3.
Figure 3. Configuration of the
Configuration of the overall
overall system.
system.
Energies 2017, 10, 166 7 of 19
Figure 3. Configuration of the overall system.

Figure
Figure 4.
4. Experimental
Experimental test
test setup.
setup.

Table 1. Specifications of a grid connected inverter.

Table 1. Specifications of a grid connected inverter.
Parameters Value Units
Parameters
Rated power Value2 kVAUnits
DC-link
Rated powervoltage 2420 VkVA
DC-link
DC-link capacitor
voltage 5500
420 μF V
DC-link capacitor
Switching frequency 550010 kHzµF
Switching frequency
Resistance (load bank) 1025 ΩkHz
Resistance (load
Filter bank)
resistance 250.5 Ω Ω
FilterFilter
resistance
inductance 0.5 7 mHΩ
Filter inductance 7 mH
Grid voltage 380 V
Grid voltage 380 V
Grid Grid frequency
frequency 6060 HzHz

Figure 5 shows various grid voltages used for test purpose, which are generated by three-phase
Figure 5 shows
programmable various
AC power grid voltages
source. Figure 5aused for testthe
represents purpose, which
ideal grid are generated
voltages in whichby
thethree-phase
frequency
programmable AC power source. Figure 5a represents the ideal grid voltages
is 60 Hz and the line-to-line voltage is 220 V in root-mean-square (RMS). Figure 5b in which the frequency
shows
is 60 Hz and the line-to-line voltage is 220 V in root-mean-square (RMS). Figure 5b shows
three-phase unbalanced grid voltages with 20% magnitude reduction in c-phase. Figure 5c shows the three-phase
unbalanceddistorted
harmonic grid voltages
grid with 20% magnitude
voltages including reduction
5% of the in c-phase. Figure
fifth- and 5c shows theharmonics,
seventh-order harmonic
distorted grid voltages including 5% of the fifth- and seventh-order harmonics, respectively.
respectively.

Energies 2017, 10, 166 8 of 18

(a) (b)

(c)
Figure 5.
Figure 5. Grid
Grid voltages
voltages under
under various
various conditions.
conditions. (a)
(a) Ideal
Ideal grid
grid voltages;
voltages; (b)
(b) three-phase
three-phase unbalanced
unbalanced
grid voltages; and (c) harmonic distorted grid voltages.
grid voltages; and (c) harmonic distorted grid voltages.

Figure 6 shows the simulation results under the ideal grid condition to illustrate the estimation
performance for the grid voltages and phase angle using DOB. The grid voltages are estimated in the
stationary frame using Equation (15) as represented in Figure 2. As is clearly observed from Figure
6a, even if the proposed scheme estimates the stationary q-axis and d-axis grid voltages well, the
estimated voltages have small phase delay from the actual grid voltage. This result was expected
because two filtered signals by the LPF were employed to obtain the estimates for the grid voltage in
Equation (15). However, this phase delay can be easily calculated. As a result, the phase-lead
 (c)
Figure 5. Grid voltages under various conditions. (a) Ideal grid voltages; (b) three-phase unbalanced
Energies 2017, 10, 166 8 of 19
grid voltages; and (c) harmonic distorted grid voltages.

Figure
Figure66shows
showsthethesimulation
simulation results under the
results under the ideal
idealgrid
gridcondition
conditiontotoillustrate
illustratethe theestimation
estimation
performance for the grid voltages and phase angle using DOB. The grid voltages
performance for the grid voltages and phase angle using DOB. The grid voltages are estimated are estimated ininthe
stationary frameframe
the stationary usingusing
Equation (15) as
Equation represented
(15) in Figure
as represented 2. As2.is As
in Figure clearly observed
is clearly from Figure
observed from
6a, even if the proposed scheme estimates the stationary q-axis and d-axis
Figure 6a, even if the proposed scheme estimates the stationary q-axis and d-axis grid voltagesgrid voltages well,
well,the
estimated voltages
the estimated havehave
voltages small
smallphase delay
phase from
delay fromthe
theactual
actualgrid
grid voltage. This result
voltage. This resultwas wasexpected
expected
because
because two filtered signals by the LPF were employed to obtain the estimates for the grid voltagein
two filtered signals by the LPF were employed to obtain the estimates for the grid voltage
Equation (15).(15).
in Equation However,
However, thisthis
phase
phasedelay
delaycan
canbebeeasily
easily calculated.
calculated. AsAs aa result,
result,thethephase-lead
phase-lead
compensation to reconstruct the
compensation to reconstruct the phase phase angle of actual grid voltage can be simply accomplished byby
actual grid voltage can be simply accomplished
Equation
Equation(18).
(18).Figure
Figure6b
6bshows
shows thethe PLL
PLL results using
using the
the estimated
estimatedgrid gridvoltages
voltageswithwithphase-lead
phase-lead
compensation.
compensation. For comparison,
comparison,two two
PLLPLL results
results are shown
are shown where
where one one is using
is obtained obtained using the
the measured
measured
voltages and voltages andusing
the other the other using the
the estimated estimated
voltages. voltages.
It is clearly It isthat
shown clearly shown thatwell
two waveforms two
waveforms
coincide with well coincide
each with each
other within other within
three periods, whichthree
provesperiods, which
the validity of proves the validity
the proposed of the
grid voltage
estimation
proposed gridscheme.
voltage estimation scheme.

(a)

(b)

Figure 6. Estimation of grid voltages and phase angle using DOB under the ideal grid condition. (a)
Figure 6. Estimation of grid voltages and phase angle using DOB under the ideal grid condition.
Measured and estimated stationary q-axis and d-axis grid voltages; and (b) phase-locked loop (PLL)
(a) Measured and estimated stationary q-axis and d-axis grid voltages; and (b) phase-locked loop (PLL)
results
resultsusing
usingthe
themeasured
measured voltages and estimated
voltages and estimatedvoltages
voltagesatatstart-up.
start-up.

Figure 7 shows the estimation performance of the grid voltages and phase angle using the
Figure 7 shows the estimation performance of the grid voltages and phase angle using the
proposed DOB based scheme when three-phase grid voltages are in unbalanced condition as shown
proposed DOB based scheme when three-phase grid voltages are in unbalanced condition as shown in
in Figure 5b. Even in this case, the DOB still works successfully except for small phase delay as
Figure 5b. Even in this case, the DOB still works successfully except for small phase delay as expected.
expected. Similarly, the exact phase angle of the grid voltages can be determined by the proposed
Similarly, the exact phase angle of the grid voltages can be determined by the proposed scheme in
scheme in a few cycles.
a few cycles.
Figure 8 shows the simulation results for the estimation performance of the grid voltages and
phase angle using the proposed scheme when the grid voltages have harmonic distortion as represented
in Figure 5c. Despite the harmonic distortion, the proposed scheme shows an excellent performance in
the estimation of the grid voltages as well as in the determination of the phase angle.
Energies 2017, 10, 166 9 of 18

Energies 2017, 10, 166 9 of 19

Energies 2017, 10, 166 9 of 18

(a)

(a)

(b)
Figure 7. Estimation of grid voltages and phase angle using DOB under unbalanced grid condition.
(a) Measured and estimated stationary q-axis and d-axis grid voltages; and (b) PLL results using
measured voltages and estimated voltages at start-up.
(b)
Figure 8 shows the simulation results for the estimation performance of the grid voltages and
phaseFigure7.7.Estimation
angle
Figure Estimation
using theof ofproposed
gridvoltages
grid voltages andphase
scheme
and phase
when angle using
theusing
angle DOB
gridDOB underhave
voltages
under unbalanced grid condition.
harmonic
unbalanced grid condition.
distortion as
(a) Measured
(a) Measured and estimated
and estimated stationary
stationary q-axis and
q-axis and d-axis grid
d-axis grid voltages; and (b) PLL results using
represented in Figure 5c. Despite the harmonic distortion, thevoltages;
proposed and (b) PLLshows
scheme resultsanusing
excellent
measured
measured voltages
voltages and
and estimated
estimated voltages
voltages at
at start-up.
start-up.
performance in the estimation of the grid voltages as well as in the determination of the phase angle.

Figure 8 shows the simulation results for the estimation performance of the grid voltages and
phase angle using the proposed scheme when the grid voltages have harmonic distortion as
represented in Figure 5c. Despite the harmonic distortion, the proposed scheme shows an excellent
performance in the estimation of the grid voltages as well as in the determination of the phase angle.

(a)

(a)

(b)

Figure8.8. Estimation
Figure Estimation of
of grid
grid voltages
voltages and
and phase
phase angle
angle using
using DOB
DOB under
under harmonic
harmonic distorted
distorted grid
grid
condition.(a)
condition. (a)Measured
Measuredand
andestimated
estimatedstationary
stationaryq-axis
q-axisand
andd-axis
d-axisgrid
gridvoltages;
voltages;and
and(b)
(b)PLL
PLLresults
results
usingmeasured
using measuredvoltages
voltagesand
andestimated
estimatedvoltages
voltagesatatstart-up.
start-up.
(b)
Toverify
To verifythe
thefeasibility
feasibilityof
ofthe
theproposed
proposedscheme,
scheme,Figure
Figure99illustrates
illustratesthe
thecomparative
comparativesimulation
simulation
Figure
resultsfor 8.
forthe
theEstimation
inverter of grid
output voltages
currents and phase
under angle
the idealusing DOB under
grid condition harmonic
between distorted grid
the conventional
results
condition.inverter output
(a) Measured currents
and under
estimated the ideal
stationary gridand
q-axis condition between
d-axis grid theand
voltages; conventional scheme
(b) PLL results
scheme
with withmeasurements
voltage voltage measurements
and the and thevoltage-sensorless
proposed proposed voltage-sensorless
control control
scheme. It is scheme.from
obvious It is
using measured voltages and estimated voltages at start-up.
these figures that the control performance of the proposed scheme is comparable to the case with
voltage
Tomeasurements.
verify the feasibility of the proposed scheme, Figure 9 illustrates the comparative simulation
results for the inverter output currents under the ideal grid condition between the conventional
scheme with voltage measurements and the proposed voltage-sensorless control scheme. It is
Energies
Energies 2017,
2017, 10,
10, 166
166 10
10 of
of 18
18

obvious
obvious
Energies from
from
2017,
Energies 2017, these
166 figures
these
10,10,
166 figures that
that the
the control
control performance
performance of
of the
the proposed
proposed scheme
scheme is
is comparable
comparable to
to
1010 the
the
ofof
1819
case
case with
with voltage
voltage measurements.
measurements.
obvious from these figures that the control performance of the proposed scheme is comparable to the
case with voltage measurements.

(a)
(a) (b)
(b)
Figure
Figure 9.
Figure 9. Comparison
9. Comparison
Comparison(a)
of inverter
ofof
inverter output
output
inverter currents
currents
output under
under
currents the
the ideal
under ideal grid
the ideal (b) (a)
grid condition.
condition. (a) Inverter
grid condition.Inverter output
output
(a) Inverter
currents
currents
output with
with
currents voltage
voltage measurements;
measurements;
with voltage and
and
measurements; (b)
(b) inverter
inverter
andunder output
(b) inverteroutput currents
currents with
with the
the proposed
proposed
Figure 9. Comparison of inverter output currents the idealoutput currents (a)
grid condition. with the proposed
Inverter output
voltage-sensorless
voltage-sensorless
voltage-sensorless control.
control. measurements; and (b) inverter output currents with the proposed
control.
currents with voltage
voltage-sensorless control.
Figure
Figure 10 10 represents
represents thethe comparative simulation results
comparative simulation results under
under three-phase unbalanced grid
three-phase unbalanced grid
voltages
voltages as
as
as in
in
in Figure
Figure
Figure 5b
5b
5b for
for
for the
the
the schemes
schemes
schemes with
with
with and
andand without
without
without thethe
the voltage
voltage
voltage sensors.
sensors.
sensors. It isIt
It is
is clearly
clearly
clearly
Figure 10 represents the comparative simulation results under three-phase unbalanced grid shownshown
shown
that
that
that
both both
both schemes
schemes
schemes
voltages yield
as in yield
yield similar
5bsimilar
similar
Figure current
thecurrent
current
for waveforms,
waveforms,
waveforms,
schemes which
which
which
with and proves
proves
proves
without that
the that
the
voltage the proposed
thatproposed
the proposed
sensors. scheme
It scheme
scheme
works
is clearly works
works
well
shown
well
well
even even
even
that in
inboth this
this condition.
inschemes
this condition.
condition. yield similar current waveforms, which proves that the proposed scheme works
well even in this condition.

(a)
(a) (b)
(b)
(a) (b)
Figure 10.
10. Comparison
Figure 10.
Figure Comparison of
Comparison of inverter
of inverter output
inverter currents
output currents
output under
currents under unbalanced
under unbalanced grid
unbalanced grid condition.
condition. (a)
grid condition. (a) Inverter
(a) Inverter
Inverter
Figure 10. Comparison of inverter output currents under unbalanced grid condition. (a) Inverter
output
output currents
output currents with
currents with voltage
with voltage measurements;
voltage measurements; and
measurements; and (b)
and (b) inverter
(b) inverter output
inverter output currents
output currents with
currents with the
with the proposed
the proposed
proposed
output currents with voltage measurements; and (b) inverter output currents with the proposed
voltage-sensorless
voltage-sensorless control.
voltage-sensorless control.
control.
voltage-sensorless control.

Figure
Figure 11 11 shows
11shows
showsthethethe comparative
comparative
comparative simulation
resultsresults
simulation
simulation results
under the under
under the
the harmonic
harmonic harmonicgrid
distorted distorted
distorted
condition grid
grid
as
Figure 11 shows the comparative simulation results under the harmonic distorted grid
condition
condition
in Figure 5cas in
asfor Figure
inthe
Figure
schemes 5c
5c for
for
withthe
the
andschemes
schemes
without with
with
the and
and without
voltage without
sensors. the
the voltage
voltage
Similar to sensors.
sensors.
the previous Similar
Similar to
to the
simulations, the
condition as in Figure 5c for the schemes with and without the voltage sensors. Similar to the
previous
previous
Figure simulations,
11a,b representsFigure
simulations, Figure
the 11a,b
11a,b
inverter represents
represents for the
the inverter
the inverter currents
currents for
for thethe conventional
the conventional
conventional and
and
previous simulations, Figure 11a,bcurrents
represents the conventional
inverter currents and forproposed scheme. On and the
proposed
proposed
other
proposed
scheme.
hand,scheme.
Figure On
On
scheme.11c
the
the
Onshows
other
theother
other
hand,
thehand,
fast
hand,
Figure
Figure
Fourier 11c
11c shows
11c shows
Figuretransformshows
the
the
(FFT) fast
fast Fourier
Fourier
result
the fast Fourier
transform
transform
for a-phase
transform
(FFT)
(FFT)
current
(FFT)
result
resultfor
waveform
result
for
for
in
a-phase
a-phase
Figure current
current
11b together waveform
waveform
with the in
in Figure
Figure
harmonic 11b
11b
limits together
together
on IEEE with
with
Standard the
the harmonic
harmonic
519-1992.
a-phase current waveform in Figure 11b together with the harmonic limits on IEEE Standard In limits
limits
this on
on
case, IEEE
IEEE
both Standard
Standard
the current
519-1992.
519-1992.
waveform
519-1992.In
In this
and
Inthis case,
case,
the
this power
case, both
both
both the
thecurrent
the
qualitycurrent waveform
waveform
are slightly
current waveform and
and the
and
degraded the
thein power
power
power quality are
qualitycontrol
the proposed
quality slightly
areslightly
are slightly
scheme degraded
degraded inin
as compared
degraded the
inthe
the
proposed
proposed
with control
control
the conventional
proposed scheme
controlscheme
scheme as
scheme compared
asascompared
due to the
compared with
with the
the conventional
the conventional
bandwidth
with scheme
scheme
limit ofscheme
conventional the LPFduedue
due
usedto
to the
tothe bandwidth
theDOB.
in bandwidth
However,
bandwidth limit
limitof
limit of
of
the
the LPF
thethe
LPF
harmonics used
used
on in
in DOB.
DOB.
inverter However,
However,
currents are the
the harmonics
harmonics
effectively on
maintainedon inverter
inverter
within currents
currents
the
LPF used in DOB. However, the harmonics on inverter currents are effectively maintained are
harmonicare effectively
effectively
limits maintained
maintained
specified on IEEE
within
within
Standardthe
thethe
within harmonic
harmonic
519-1992 aslimits
harmonic limits
inlimits specified
Figure 11c. on
specified
specified ononIEEE
IEEEStandard
IEEE Standard 519-1992
Standard 519-1992 as in
as in
as Figure
in Figure
Figure11c.11c.
11c.

(a)
(a)
(a) (b)
(b)
(b)

Figure 11. Cont.

Energies 2017, 10, 166 11 of 19
Energies 2017, 10, 166 11 of 18
Energies 2017, 10, 166 11 of 18

(c)
(c)

FigureFigure
Figure11.11.11. Comparisonof
Comparison
Comparison ofinverter
of inverter output
inverter output
outputcurrents
currents under
currentsunder harmonic
underharmonic distorted
harmonic distortedgrid condition.
grid
distorted (a) (a)
condition.
grid condition.
Inverter
Inverter output
output currentswith
currents withvoltage
voltage measurements;
measurements; (b)(b)inverter
inverteroutput
output currents with
currents thethe
with proposed
proposed
(a) Inverter output currents with voltage measurements; (b) inverter output currents with the proposed
voltage-sensorlesscontrol;
voltage-sensorless control;and
and(c)
(c)fast
fast Fourier
Fourier transform
transform(FFT)
(FFT)result
resultfor a-phase
for a-phasecurrent in (b).
voltage-sensorless control; and (c) fast Fourier transform (FFT) result for a-phase current
currentinin(b).
(b).
Figure1212shows
Figure showsthethe estimation
estimation performance
performance ofof grid
gridvoltages
voltages and
andcorresponding
corresponding inverter
inverter
Figure
output 12 shows
current the estimation
waveforms under performance of grid
different inductance voltages
variations. and
Evencorresponding inverter
though the magnitude output
of
output
current current waveforms under different inductance variations. Even though the magnitude of
current ripple is varied due to the different inductance values, the estimation performance of gridripple
waveforms under different inductance variations. Even though the magnitude of current
current ripple is varied due to the different inductance values, the estimation performance of grid
is varied due istonot
voltages theaffected
differentbyinductance values,which
such a variation, the estimation
well provesperformance of of
the robustness grid
thevoltages
proposedis not
voltages is not affected by such a variation, which well proves the robustness of the proposed
affected by suchalgorithm.
estimation a variation, which well proves the robustness of the proposed estimation algorithm.
estimation algorithm.

(a) (b)

(a) (b)

(c)
Figure 12. Estimation of grid voltages and inverter output currents under inductance variation.
(a) +20% variation; (b) +10% variation; and (c) −20% variation.
(c)
Energies 2017, 10, 166 12 of 18

Figure 12. Estimation of grid voltages and inverter output currents under inductance variation. (a)
Energies 2017, 10, 166 12 of 19
+20% variation; (b) +10% variation; and (c) −20% variation.

Figure
Figure1313shows
showsthe
thefrequency
frequencyresponses
responses for for the open-loop transfer
transfer function
function of
ofFigure
Figure22when
when
the voltage
the voltagefeed-forward
feed-forwardisisconsidered
considered under
under the
the nominal value
value of
of inductance
inductanceandand+20%
+20%variation
variationofof
inductance,respectively.
inductance, respectively. Despite
Despite the
theinductance
inductancevariation,
variation,thethe
frequency responses
frequency remain
responses similar
remain and
similar
both
and thethe
both frequency responses
frequency satisfy
responses the stability
satisfy criterion.
the stability In addition,
criterion. the phase
In addition, compensation
the phase does
compensation
not not
does affect the system
affect stability
the system because
stability it is only
because used
it is onlyto used
changeto the phasethe
change angle,
phaseandangle,
thus, the
andreference
thus, the
current in
reference the stationary
current frame. frame.
in the stationary

(a)

(b)
Figure
Figure13.
13.Frequency
Frequency responses forfor
responses thethe
open-loop transfer
open-loop function
transfer under
function inductance
under variation.
inductance (a)
variation.
nominal valuevalue
(a) Nominal of inductance; and and
of inductance; (b) +20% variation.
(b) +20% variation.

Figure
Figure1414shows
showsthe theestimation
estimation performance
performance ofof grid voltages using
grid voltages using the
the conventional
conventionallinear
linear
Luenberger type disturbance observer under the ideal grid voltages, unbalanced grid
Luenberger type disturbance observer under the ideal grid voltages, unbalanced grid voltages, voltages, and
harmonic distorted
and harmonic grid voltages,
distorted respectively.
grid voltages, The estimation
respectively. performance
The estimation of gridof
performance voltages is more
grid voltages
orisless similar to that of the proposed scheme depending on the observer error dynamics.
more or less similar to that of the proposed scheme depending on the observer error dynamics. This
scheme also has
This scheme alsothe
haslimitation thatthat
the limitation sinusoidally varying
sinusoidally signals
varying signalshave
havetotobe
be tracked, and, thus,
tracked, and, thus,the
the
current control performance will be much degraded without proper phase compensation.
current control performance will be much degraded without proper phase compensation.
To validate the actual availability of the proposed scheme, the experiments are carried out using
the equipment depicted in Figures 3 and 4.
 Energies 2017, 10, 166 13 of 18
Energies 2017, 10, 166 13 of 19
Energies 2017, 10, 166 13 of 18

(a)

(a)

(b)

(b)

(c)

Figure 14. Estimation of grid voltages using the conventional linear Luenberger type disturbance
observer. (a) Under the ideal grid voltages; (b) (c)under unbalanced grid voltages; and (c) under
harmonic
Figure distorted grid
14. Estimation of voltages.
grid voltages using the conventional linear Luenberger type disturbance
Figure 14. Estimation of grid voltages using the conventional linear Luenberger type disturbance
observer.
observer. (a) Under
(a) Underthethe ideal
ideal gridgrid voltages;
voltages; (b) under unbalanced grid voltages; and harmonic
(c) under
To validate
harmonic the actual
distorted grid availability
voltages. of(b)the
under unbalanced
proposed grid voltages;
scheme, and (c) under
the experiments are carried out
distorted grid voltages.
using the equipment depicted in Figures 3 and 4.
ToFigure 15 shows
validate the experimental
the actual availability ofresults for the DOB
the proposed basedthe
scheme, estimating performance
experiments of grid
are carried out
Figure
voltages 15 shows
under threethedifferent
experimental
grid results for the
conditions as DOB based estimating
presented in Figure performance
5. The of grid voltages
experimental results
using the equipment depicted in Figures 3 and 4.
under
accordthree different
well15
with thegrid conditionsresults
simulation as presented in Figure
in Figures 5. The experimental
6–8,DOB
yielding results information
accord well witha
Figure shows the experimental results for the basedthe grid voltage
estimating performance of in grid
the simulation
reasonable results
accuracy. in Figures 6–8, yielding the grid voltage information in a reasonable accuracy.
voltages under three different grid conditions as presented in Figure 5. The experimental results
accord well with the simulation results in Figures 6–8, yielding the grid voltage information in a
reasonable accuracy.

(a)

Figure 15. Cont.

(a)
Energies 2017, 10, 166 14 of 19
Energies 2017, 10, 166 14 of 18

(b)

(c)
Figure 15. Experimental results for the estimation of grid voltages using DOB. (a) Under the ideal
Figure 15. Experimental
grid voltages; (b) under results for thegrid
unbalanced estimation ofand
voltages; grid(c)
voltages
under using DOB.
harmonic (a) Under
distorted thevoltages.
grid ideal grid
voltages; (b) under unbalanced grid voltages; and (c) under harmonic distorted grid voltages.
Figure 16 shows the experimental results for the phase angle estimation and transient current
Figure of
responses 16 the
shows the experimental
proposed results forcontrol
voltage-sensorless the phase angleatestimation
scheme start-up and
undertransient current
the ideal grid
responses
voltages inof Figure
the proposed
5a. In voltage-sensorless
the initial stage, thecontrol scheme
current at start-up
references are under
set to the ideal
small grid(0.5
value voltages
A) to
in Figurelarge
prevent 5a. Inabrupt
the initial stage,
currents the current
caused by thereferences
inaccurateare set angle
phase to small value (0.5in
information A)the
to absence
prevent oflarge
the
abrupt currents
grid voltage caused by
estimation. the inaccurate
Figure phasethe
16a represents angle information
waveforms in the
of the absencea-phase
measured of the grid
gridvoltage
voltage
estimation.
and phase Figure 16a represents
angle estimation. In this figure, θ denotes
the waveforms the phasea-phase
of the measured grid voltage
angle obtained usingandthe
phase angle
measured
estimation. In thisisfigure,
voltages, which referredθ denotes the phase angle obtained
as the real using the
of grid voltages measured
and used forvoltages, which is
only comparison
referred as the real phase angle of grid voltages and used for only comparison purpose. On the other
purpose. On the other hand, θ̂ c denotes the phase angle obtained using the estimated voltages with
hand, θ̂c denotes the phase angle obtained using the estimated voltages with phase lead compensation,
phase lead compensation, which is the phase angle determined by the proposed scheme. While the
which is the phase angle determined by the proposed scheme. While the connection signal for the
connection signal for the MC is given with the “control start signal” as shown in Figure 16a, it
MC is given with the “control start signal” as shown in Figure 16a, it requires about 20 ms to operate.
requires about 20 ms to operate. The connection instant is indicated by the “grid connection by MC
The connection instant is indicated by the “grid connection by MC on”. As soon as the MC is closed
on”. As soon as the MC is closed and the inverter is connected to grid, the DOB estimates the grid
and the inverter is connected to grid, the DOB estimates the grid voltages properly. As a result, the
voltages properly. As a result, the proposed voltage-sensorless scheme detects the exact phase angle
proposed voltage-sensorless scheme detects the exact phase angle of grid voltages within two cycles.
of grid voltages within two cycles. Figure 16b represents three-phase inverter current waveforms
Figure 16b represents three-phase inverter current waveforms with the “control start signal” during
with the “control start signal” during the start-up process. Similarly, it takes about 20 ms for MC to
the start-up process. Similarly, it takes about 20 ms for MC to connect. As soon as MC is connected,
connect. As soon as MC is connected, the inverter is in grid connected mode. From that instant, both
the inverter is in grid connected mode. From that instant, both the control input and system output
the control input and system output signals are available for the DOB for the estimation of grid
signals are available for the DOB for the estimation of grid voltage and phase angle. Even though
voltage and phase angle. Even though some oscillatory currents are observed due to dynamic
some oscillatory currents are observed due to dynamic characteristics of the DOB after MC connection
characteristics of the DOB after MC connection to grid, such a transient behavior disappears in few
to grid, such a transient behavior disappears in few milliseconds. As a result, despite the uncertainty
milliseconds. As a result, despite the uncertainty due to the absence of voltage sensors, the proposed
due to the absence of voltage sensors, the proposed scheme stabilizes the system operation rapidly
scheme stabilizes the system operation rapidly without giving a harmful damage to system.
without giving a harmful damage to system.
Figure 17 shows the comparative experimental results for the schemes with and without the
voltage sensors when the grid voltages are in ideal condition as shown in Figure 5a. The grid voltages
are estimated in the stationary frame as (15). These estimated grid voltages are used to generate the
phase angle through the SRF-PLL with phase lead compensation as explained in Figure 2. As expected,
the current waveforms are quite similar for both the cases, which is well coincident with the simulation
results in Figure 9. This confirms the effectiveness of the proposed voltage-sensorless control scheme
in an experimental way.

(a)
connect. As soon as MC is connected, the inverter is in grid connected mode. From that instant, both
the control input and system output signals are available for the DOB for the estimation of grid
voltage and phase angle. Even though some oscillatory currents are observed due to dynamic
characteristics of the DOB after MC connection to grid, such a transient behavior disappears in few
milliseconds.
Energies 2017, 10,As
166a result, despite the uncertainty due to the absence of voltage sensors, the proposed
15 of 19
scheme stabilizes the system operation rapidly without giving a harmful damage to system.

Energies 2017, 10, 166 15 of 18

Energies 2017, 10, 166 15 of 18

(a)

(b)
Figure 16. Phase angle estimation and transient current responses of the proposed voltage-sensorless
control scheme at start-up. (a) Waveforms of the measured a-phase grid voltage and phase angle
estimation ( θ : PLL result using the measured voltages, θ̂c : PLL result using the estimated voltages
with compensation); and (b) inverter output currents at start-up.

Figure 17 shows the comparative experimental results for the schemes with and without the
voltage sensors when the grid voltages are in (b) ideal condition as shown in Figure 5a. The grid
voltages are estimated in the stationary frame as (15). These estimated grid voltages are used to
Figure
Figure
generate 16.phase
16.
the Phaseangle
angle estimation
through the andSRF-PLL current
transientwith
current responses
responses
phase of the
of the proposed
proposedas
lead compensation voltage-sensorless
voltage-sensorless
explained in Figure
control scheme
scheme at at start-up.
start-up. (a) Waveforms of the the measured
measured a-phase
a-phase grid
grid voltage
voltage and
and phase
control
2. As expected, the current waveforms are quite similar for both the cases, which is phase angle
angle
well coincident
estimation
with estimation ( θ : PLL result
(θ: PLL
the simulation result using
using
results the measured
in the measured
Figure voltages,
9. This θ̂
voltages,c θ̂ : PLLthe
: PLL
confirms
c
result using the estimated
resulteffectiveness voltages
using the estimated
of the with
voltages
proposed
compensation); and (b) inverter output currents at start-up.
voltage-sensorless control
with compensation); andscheme in anoutput
(b) inverter experimental
currents atway.
start-up.

Figure 17 shows the comparative experimental results for the schemes with and without the
voltage sensors when the grid voltages are in ideal condition as shown in Figure 5a. The grid
voltages are estimated in the stationary frame as (15). These estimated grid voltages are used to
generate the phase angle through the SRF-PLL with phase lead compensation as explained in Figure
2. As expected, the current waveforms are quite similar for both the cases, which is well coincident
with the simulation results in Figure 9. This confirms the effectiveness of the proposed
voltage-sensorless control scheme in an experimental way.

(a)

(a)
(b)
Figure 17.
Figure 17. Comparative
Comparative experimental
experimental results for inverter output currents under
under the
the ideal
ideal grid
grid
condition. (a) With the conventional PR control with voltage measurements; and (b) with
condition. (a) With the conventional PR control with voltage measurements; and (b) with the proposedthe
proposed voltage-sensorless
voltage-sensorless control. control.

Figure 18 represents the experimental results for the proposed voltage-sensorless control
scheme of a grid connected inverter when the grid voltages are unbalanced or harmonic distorted as
shown in Figure 5b,c. Figure 18a,b represents three-phase inverter current waveforms under
(b)
unbalanced grid voltages and under harmonic distorted grid voltages, respectively. Generally, the
harmonics on grid voltages directly influence on the current
Figure 17. Comparative experimental results for inverter output control performance,
currents reducing
under the ideal grid the
power quality (a)
condition. of DG
Withsystem. Even though
the conventional PR the current
control with waveforms are slightlyand
voltage measurements; distorted due
(b) with theto the
proposed voltage-sensorless control.
Energies 2017, 10, 166 16 of 19

Figure 18 represents the experimental results for the proposed voltage-sensorless control scheme
of a grid connected inverter when the grid voltages are unbalanced or harmonic distorted as shown
in Figure 5b,c. Figure 18a,b represents three-phase inverter current waveforms under unbalanced
grid voltages and under harmonic distorted grid voltages, respectively. Generally, the harmonics on
grid voltages directly influence on the current control performance, reducing the power quality of DG
Energies 2017, 10, 166 16 of 18
system. Even though the current waveforms are slightly distorted due to the harmonic distortion in
grid voltages in Figure 18b, the current harmonics are within the limits specified on IEEE Standard
harmonic distortion in grid voltages in Figure 18b, the current harmonics are within the limits
519-1992 as illustrated in Figure 18c, which proves that the proposed voltage-sensorless scheme still
specified on IEEE Standard 519-1992 as illustrated in Figure 18c, which proves that the proposed
works even under the harmonic distorted grid.
voltage-sensorless scheme still works even under the harmonic distorted grid.

(a)

(b)

(c)
Figure 18. Experimental
Figure 18. Experimentalresults
resultsforfor inverter
inverter output
output currents
currents of theofproposed
the proposed voltage-sensorless
voltage-sensorless control
control
scheme under distorted grid condition. (a) Under unbalanced grid voltages; (b) under (b)
scheme under distorted grid condition. (a) Under unbalanced grid voltages; under
harmonic
harmonic distorted
distorted grid gridand
voltages; voltages;
(c) FFTand (c) FFT
result result for
for a-phase a-phase
current current in (b).
in (b).

To demonstrate the transient performance of the proposed scheme, Figure 19 shows the
To demonstrate the transient performance of the proposed scheme, Figure 19 shows the
experimental results under the step change in the q-axis current reference from 1.5 A to 3 A. Figure
experimental results under the step change in the q-axis current reference from 1.5 A to 3 A. Figure 19a
19a shows a-phase and b-phase currents, and Figure 19b shows the q-axis current. As shown in these
shows a-phase and b-phase currents, and Figure 19b shows the q-axis current. As shown in these figures,
figures, despite a slight oscillation depending on the choice of resonant gains, the currents are
despite a slight oscillation depending on the choice of resonant gains, the currents are controlled to the
controlled to the reference immediately. From these results, it is confirmed that the proposed scheme
reference immediately. From these results, it is confirmed that the proposed scheme operates stably
operates stably during transient periods.
during transient periods.

(a)
 To demonstrate the transient performance of the proposed scheme, Figure 19 shows the
experimental results under the step change in the q-axis current reference from 1.5 A to 3 A. Figure
19a shows a-phase and b-phase currents, and Figure 19b shows the q-axis current. As shown in these
figures, despite a slight oscillation depending on the choice of resonant gains, the currents are
controlled
Energies to the
2017, 10, 166 reference immediately. From these results, it is confirmed that the proposed scheme
17 of 19
operates stably during transient periods.

Energies 2017, 10, 166 17 of 18

(a)

(b)
Figure19.
Figure 19. Experimental
Experimental results
results of
of the
the proposed
proposed control
control scheme
scheme under
under the
the step
stepchange
changein
incurrent
current
reference. (a) a-phase and b-phase current responses; and (b) q-axis current response.
reference. (a) a-phase and b-phase current responses; and (b) q-axis current response.

6. Conclusions
6. Conclusions
To enhance the economic feasibility in DG systems without the loss of reliability, this paper
To enhance the economic feasibility in DG systems without the loss of reliability, this paper
presents a voltage-sensorless control scheme for a grid connected inverter using DOB. A grid
presents a voltage-sensorless control scheme for a grid connected inverter using DOB. A grid connected
connected inverter usually requires voltage measurements to detect the correct phase angle of the
inverter usually requires voltage measurements to detect the correct phase angle of the grid voltages.
grid voltages. To obtain the grid voltage information without using the voltage sensors, the
To obtain the grid voltage information without using the voltage sensors, the proposed scheme employs
proposed scheme employs a DOB based technique to estimate the grid voltage from the current
a DOB based technique to estimate the grid voltage from the current measurements and reference
measurements and reference signals. By using the proposed DOB based estimation with a phase
signals. By using the proposed DOB based estimation with a phase lead compensation, the phase
lead compensation, the phase angle of grid voltages can be completely restored even if the phase
angle of grid voltages can be completely restored even if the phase angle of grid is initially unknown.
angle of grid is initially unknown. The proposed scheme mainly consists of two parts that are a
The proposed scheme mainly consists of two parts that are a control design of grid connected inverter
control design of grid connected inverter based on the PR controller and the estimator design for the
based on the PR controller and the estimator design for the grid voltages as well as the phase angle.
grid voltages as well as the phase angle. The proposed scheme is simple and straightforward. In
The proposed scheme is simple and straightforward. In addition, it does not require any additional
addition, it does not require any additional hardware. Even in the initial stage, the proposed scheme
hardware. Even in the initial stage, the proposed scheme does not cause any damage to system due
does not cause any damage to system due to the uncertainty in phase angle. To verify the feasibility
to the uncertainty in phase angle. To verify the feasibility of the proposed voltage-sensorless control
of the proposed voltage-sensorless control scheme, a 2 kVA prototype grid connected inverter has
scheme, a 2 kVA prototype grid connected inverter has been constructed using the DSP TMS320F28335.
been constructed using the DSP TMS320F28335. Through the comparative simulations and
Through the comparative simulations and experiments, it has been validated that the proposed scheme
experiments, it has been validated that the proposed scheme works effectively even under the
works effectively even under the uncertain grid such as the imbalance and harmonic distortion.
uncertain grid such as the imbalance and harmonic distortion.
Acknowledgments: This research was supported by Basic Science Research Program through the National
Acknowledgments:
Research FoundationThis research
of Korea wasfunded
(NRF) supported by Basic
by the Science
Ministry Research (NRF-2016R1D1A1B03930975).
of Education Program through the National
This work was
Research also supported
Foundation of Koreaby(NRF)
the Human
funded Resources Development
by the Ministry of the Korea
of Education Institute of Energy Technology
(NRF-2016R1D1A1B03930975). This
Evaluation and Planning (KETEP) grant funded by the Korea government Ministry
work was also supported by the Human Resources Development of the Korea Institute of Trade, IndustryTechnology
of Energy & Energy
(NO. 20154030200720).
Evaluation and Planning (KETEP) grant funded by the Korea government Ministry of Trade, Industry & Energy
Author Contributions: Hyun-Sou Kim and Kyeong-Hwa Kim conceived the main concept of the control structure
(NO. 20154030200720).
and developed the entire system. Hyun-Sou Kim carried out the research and analyzed the numerical data with the
Author Contributions:
guidance from Kyeong-Hwa Hyun-Sou Kim and
Kim. Hyun-Sou KimKyeong-Hwa
and Kyeong-HwaKim conceived the main
Kim collaborated conceptthe
to prepare of manuscript.
the control
structureof
Conflicts and developed
Interest: the entire
The authors system.
declare noHyun-Sou
conflict of Kim carried out the research and analyzed the numerical
interest.
data with the guidance from Kyeong-Hwa Kim. Hyun-Sou Kim and Kyeong-Hwa Kim collaborated to prepare
the manuscript.

Conflicts of Interest: The authors declare no conflict of interest.

References
1. Mohamed, Y.A.R.I.; Rahman, M.A.; Seethpathy, R. Robust line-voltage sensorless control and
synchronization of LCL-filtered distributed generation inverters for high power quality grid connection.
IEEE Trans. Power Electron. 2012, 27, 87–98.
Energies 2017, 10, 166 18 of 19

References
1. Mohamed, Y.A.R.I.; Rahman, M.A.; Seethpathy, R. Robust line-voltage sensorless control and synchronization
of LCL-filtered distributed generation inverters for high power quality grid connection. IEEE Trans.
Power Electron. 2012, 27, 87–98. [CrossRef]
2. Judewicz, M.G.; Gonzalez, S.A.; Echeverria, N.I.; Fischer, J.R.; Carrica, D. Generalized predictive current
control (GPCC) for grid-tie three-phase inverters. IEEE Trans. Ind. Electron. 2016, 63, 4475–4484. [CrossRef]
3. Wu, W.; Sun, Y.; Huang, M.; Wang, X.; Wang, H.; Blaabjerg, F.; Liserre, M.; Chung, H.S.H. A robust passive
damping method for LLCL-filter-based grid-tied inverters to minimize the effect of grid harmonic voltages.
IEEE Trans. Power. Electron. 2014, 29, 3279–3289. [CrossRef]
4. Liserre, M.; Cardenas, R.; Molinas, M.; Rodriguez, J. Overview of multi-MW wind turbines and wind parks.
IEEE Trans. Ind. Electron. 2011, 58, 1081–1095. [CrossRef]
5. Wang, J.; Xu, D.; Wu, B.; Luo, Z. A low-cost rectifier topology for variable-speed high-power PMSG wind
turbines. IEEE Trans. Power Electron. 2011, 26, 2192–2200. [CrossRef]
6. Sosa, J.L.; Castilla, M.; Miret, J.; Matas, J.; Al-Turki, Y.A. Control strategy to maximize the power capability
of PV three-phase inverters during voltage sags. IEEE Trans. Power Electron. 2016, 31, 3314–3323. [CrossRef]
7. Lin, F.J.; Lu, K.C.; Ke, T.H.; Yang, B.H.; Chang, Y.R. Reactive power control of three-phase grid-connected PV
system during grid faults using Takagi–Sugeno–Kang probabilistic fuzzy neural network control. IEEE Trans.
Ind. Electron. 2015, 62, 5516–5528. [CrossRef]
8. Yi, H.; Zhuo, F.; Wang, F.; Wang, Z. A digital hysteresis current controller for three-level neural-point-clamped
inverter with mixed-levels and prediction-based sampling. IEEE Trans. Power Electron. 2016, 31, 3945–3957.
[CrossRef]
9. Singaravel, M.M.R.; Daniel, S.A. MPPT with single DC–DC converter and inverter for grid-connected hybrid
wind-driven PMSG–PV system. IEEE Trans. Ind. Electron. 2015, 62, 4849–4857. [CrossRef]
10. Wandhare, R.G.; Agarwal, V. Novel integration of a PV-wind energy system with enhanced efficiency.
IEEE Trans. Power Electron. 2015, 30, 3638–3649. [CrossRef]
11. Kang, S.W.; Kim, K.H. Sliding mode harmonic compensation strategy for power quality improvement of
a grid-connected inverter under distorted grid condition. IET Power Electron. 2015, 8, 1461–1472. [CrossRef]
12. Trinh, Q.N.; Lee, H.H. An advanced current control strategy for three-phase shunt active power filters.
IEEE Trans. Ind. Electron. 2013, 60, 5400–5410. [CrossRef]
13. Liserre, M.; Teodorescu, R.; Blaabjerg, F. Multiple harmonics control for three-phase grid converter systems
with the use of PI-RES current controller in a rotating frame. IEEE Trans. Power Electron. 2006, 21, 836–841.
[CrossRef]
14. Zeng, Z.; Yang, H.; Tang, S.; Zhao, R. Objective-oriented power quality compensation of multifunctional
grid-tied inverters and its application in microgrids. IEEE Trans. Power Electron. 2015, 30, 1255–1265.
[CrossRef]
15. Javadi, A.; Al-Haddad, K. A single-phase active device for power quality improvement of electrified
transportation. IEEE Trans. Ind. Electron. 2015, 62, 3033–3041. [CrossRef]
16. Lascu, C.; Asiminoaei, L.; Boldea, I.; Blaabjerg, F. High performance current controller for selective harmonic
compensation in active power filters. IEEE Trans. Power Electron. 2007, 22, 1826–1835. [CrossRef]
17. Fischer, J.R.; Gonzalez, S.A.; Carugati, I.; Herran, M.A.; Judewicz, M.G.; Carrica, D.O. Robust predictive
control of grid-tied converters based on direct power control. IEEE Trans. Power Electron. 2014, 29, 5634–5643.
[CrossRef]
18. Garcia-Cerrada, A.; Pinzon-Ardila, O.; Feliu-Batlle, V.; Roncero-Sanchez, P.; García-González, P. Application
of a repetitive controller for a three-phase active power filter. IEEE Trans. Power Electron. 2007, 22, 237–246.
[CrossRef]
19. Hosein, G.K.; Mohammad, M. Novel grid voltage estimation by means of the Newton–Raphson optimisation
for three-phase grid connected voltage source converters. IET Power Electron. 2014, 7, 2945–2953.
20. Monfared, M.; Sanatkar, M.; Golestan, S. Direct active and reactive power control of single-phase grid-tie
converters. IET Power Electron. 2012, 8, 1544–1550. [CrossRef]
21. Noguchi, T.; Tomiki, H.; Kondo, S.; Takahashi, I. Direct power control of PWM converter without
power-source voltage sensors. IEEE Trans. Ind. Appl. 1998, 34, 473–479. [CrossRef]
Energies 2017, 10, 166 19 of 19

22. Leon, A.E.; Solsona, J.A.; Valla, M.I. Control strategy for hardware simplification of voltage source
converter-based power applications. IET Power Electron. 2011, 4, 39–50. [CrossRef]
23. Hoffmann, N.; Fuchs, F.W. Minimal invasive equivalent grid impedance estimation in inductive–resistive
power networks using extended Kalman filter. IEEE Trans. Power Electron. 2014, 29, 631–641. [CrossRef]
24. Chen, W.H. Disturbance observer based control for nonlinear systems. IEEE/ASME Trans. Mechatron. 2004, 9,
706–710. [CrossRef]
25. Texas Instruments. TMS320F28335 Digital Signal Controller (DSC)—Data Manual; Texas Instruments: Dallas,
TX, USA, 2008.
26. Back, J.; Shim, H. An inner-loop controller guaranteeing robust transient performance for uncertain MIMO
nonlinear systems. IEEE Trans. Autom. Control 2009, 54, 1601–1607. [CrossRef]
27. Institute of Electrical and Electronics Engineers. Recommended Practice for Harmonic Control in Electric Power
Systems; IEEE Standard 519-1992; Institute of Electrical and Electronics Engineers: New York, NY, USA, 1993.

© 2017 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).