2010 IREP Symposium – Bulk Power System Dynamics and Control – VIII (IREP), August 1-6, 2010, Buzios,

RJ, Brazil

Dynamic Modeling of Inverter-Based Distributed Generators with Voltage Positive

Feedback Anti-Islanding Protection

Tiago R. Ricciardi, Diogo Salles, Walmir Freitas Xiaoyu Wang

Department of Electrical Energy Systems Department of Electrical Engineering
University of Campinas Tsinghua University
Brazil China

Abstract One of the most challenging aspects of designing electric

power distribution systems in the presence of distributed
This paper presents three models for dynamic analysis of generation is the need to provide protection against islanding
inverter-based distributed generators equipped with voltage [1], [3]–[5]. The ability to detect when DGs become islanded
positive-feedback anti-islanding protection. The models from the main source, feeding local loads, enable them to be
analyzed are: the switched model, where the switches are quickly disconnected from the utility, guaranteeing that the
represented in details, the average model, where the switches network can be safely and orderly restored. IBDGs can sustain
and signal generators are replaced by controlled voltage electric islands, if sufficient energy is available in the primer
sources, and the small-signal model, where the average model side, because of the fast response of IGBTs and MOSFETs, the
is linearized. This small-signal model can be used for design most commonly semiconductor power switches used in grid-
purposes. The small-signal model is confronted with the connected inverter applications. Thus, anti-islanding positive
nonlinear average model through time domain simulations and feedback based schemes specifically developed for IBDGs are
presents accurate dynamic behavior. The model introduced is introduced by the power industry and researches to improve
suitable for computational analysis of stability impact of islanding detection of IBDGs [5]–[7].
positive feedback controls, dynamic interactions between
multiple controls, non-detection zones evaluation and other Compared with the passive schemes and other active schemes,
grid connection studies in distribution systems with this kind of these techniques have a better performance on islanding
distributed generation. protection of grid-connected DGs. However, a positive
feedback path inside inverter control reduces its stability
Keywords margins and changes its dynamic behavior. Consequently, if
the positive feedback gain is too high, the distributed
Distributed generation, inverter, islanding, positive feedback, generation system may become unstable even when it is
small-signal model, stability analysis. connected to the main supply system. Moreover, the anti-
islanding scheme may limit the amount of power that can be
supplied by the DG and the penetration level of multiple
Introduction IBDGs in some feeders [7]–[9].
The interest in the connection of generators directly in electric This paper presents a linearized model of an IBDG with
power distribution networks – a fact known as distributed voltage positive feedback anti-islanding control suitable to
generation (DG) – is a worldwide phenomenon that has small-signal stability analysis. The proposed model is validated
increased considerably in recent years. Although most of the through time domain dynamic simulations of a nonlinear model
presently distributed generation plants uses rotating electrical implemented in SimPowerSystems™/Matlab®. Through the
machines, driven by steam, water or wind turbines, it is eigenanalysis of this state-space small-signal model, one can
expected that advances in solar and fuel cells technologies evaluate the impact of parameters values such as load level,
along with environmental concerns will increase the use of distribution line impedance or positive-feedback gain over the
these energy sources in electrical power systems in the near system small-signal stability and the non-detection zone (NDZ)
future [1], [2]. of the anti-islanding protection.
These emerging technologies have in common the fact that
their energy conversion process is such that the energy Principles of Positive Feedback Anti-Islanding
available in the electrical side is in direct current (DC) format. Schemes
As the distribution systems worldwide operates in alternate
current (AC), those energy sources must be interfaced by These schemes use the deviations of frequency and voltage
inverters. Inverters are static power converters that condition from normal values as positive feedback signals to influence
DC to AC. This kind of DG is commonly referred as Inverter- the operation of the DG. The hypothesis is while grid-
Based Distributed Generator (IBDG). connected, this force cannot modify the magnitude or
frequency of terminal voltage, imposed by the system. In case

978-1-4244-7467-7/10/$26.00 ©2010 IEEE

of any kind of topological changes in the distribution network se
es pon
that disconnects the IBDG from the bulk power system, small r ter R
disturbances reflected on voltage magnitude or frequency at the q Inve q
point of common coupling (PCC) by power mismatching
between load and generation will be amplified in such form
that islanding detection time reduces compared with another
approaches. d d

However, the positive feedback schemes always attempt to

destabilize a generator no matter if it is islanded or not. The tive
Posi ack
assumption of stiffness in the terminal voltage magnitude or b
frequency may not be true if the DG penetration level is high, Feed
Fig. 2. GEVS scheme 1 dq diagram implementation.
the positive feedback gain is enough high or if multiple IBDGs
are connected in a same feeder [8]–[12]. The frequency based
methods modeling is previously presented [7],[11] and the Distribution Systems Modeling in the Presence of
General Electric Voltage Scheme (GEVS) is selected for IBDGs
investigation in this paper.
This section develops the model of the inverter, its associated
Voltage Based Methods: GEVS control systems, anti-islanding schemes, the distribution
network and loads. The model described by a set of nonlinear
The voltage-based schemes like General Electric Voltage equations is linearized around one operation point and then
Scheme (GEVS) inserts a small perturbation in active power derived in a generalized state space useful for small-signal
control based on the deviation of magnitude of terminal voltage stability analysis.
from its nominal value [5]–[7]. It is sketched in Fig. 1.
Inverter Model
A positive feedback path between sensed terminal voltage V
and the active power control commands an active power There are a lot of inverter topologies and control strategies
injection variation in the same direction of the ∆V error described in the technical literature. The most commonly used
between V sensed and its nominal value. Due to load in DG applications and, therefore, selected to be investigated in
characteristics, this active power injection variation ∆P will this work is the three-phase voltage source inverter (VSI). In
again imply in a voltage variation in the same direction, closing this paper is considered only the constant current injection
the positive loop. As in the frequency-based schemes, is inverter, because this is the control most used in photovoltaic
desired that this effect only can really change the terminal inverter applications.
voltage V when inverter operates in an islanded condition.
Switching Model. The first approach in modeling inverters is
The General Electric Anti-Islanding Schemes were the first to represent the device in the level of switching of the power
introduce the dq implementation of a positive feedback control. electronics elements such as IGBTs and MOSFETs, the
According to its inventors, when introduced, these methods discrete control and modulation circuitry that triggers the
have a better performance compared to the existing ones [7]. switch conduction/blockade, the harmonic filters, the DC
Although GEVS is actually a group of schemes based in the source, etc, as shown in Fig. 3. This approach is not
voltage positive feedback control, this paper presents only the appropriate to a state space representation and moreover, it is
modeling of GEVS scheme 1, as described in [7], between computationally expensive because the simulation step time
inverter terminal direct axis voltage vd and direct axis current should be tiny for a good representation of the switch state
reference idref as shown in Fig. 2. on/off transition.

Average Model. The average modeling technique is more

suitable to modal stability analysis. In this approach, the three-
phase inverter is changed by a three-phase controlled voltage
source, as shown in Fig. 4. The control system behind this
voltage source is realized through linear control strategies and
its inputs are current (power) reference for the case of current
(power) constant injection controlled IBDG.
Fig. 1. General voltage-based positive feedback anti-islanding principle of
operation. The dynamic equations of this electric circuits plus linear
control theory approach are described in the synchronous
reference frame dq. If the direct axis is aligned with phase A of
the terminal inverter voltage, then the active power injection
control is decoupled from the reactive injection power control.
The transformation between the three-phase reference frame
abc and the synchronous reference frame dq is presented in (1).

⎡ ⎛ 2 ⎞ ⎛ 2 ⎞⎤
sin θ sin⎜ θ − π ⎟ sin ⎜ θ + π ⎟⎥ ⎡va ⎤
⎡vd ⎤ 2 ⎢ ⎝ 3 ⎠ ⎝ 3 ⎠ ⎢ ⎥
⎢v ⎥ = ⎢ ⎥ ⋅ vb
⎢ ⎥ (1)
⎣ q ⎦ 3 ⎢cos θ cos⎛⎜ θ − 2 π ⎞⎟ cos⎛⎜ θ + 2 π ⎞⎟⎥ ⎢v ⎥
⎢⎣ ⎝ 3 ⎠ ⎝ 3 ⎠⎦ ⎣ c ⎦
⎥

The angle θ used in this decomposition must be tracked by a

phase-locked loop (PLL) circuit, as shown in Fig. 5. The PLL Fig. 5. PLL structure.
provides the inverter control with angle and frequency
reference.
The Kirchhoff’s Voltage Law (KVL) applied to the mesh
⎛ k ⎞ formed by the DG terminal voltage, inverter filter and the
ω PLL = ⎜ k pPLL + iPLL ⎟vq + ω0 (2)
⎝ s ⎠ controlled source leads to:
dθ PLL d
ω PLL = = (ω0t + δ PLL ) (3)
dt dt ⎡vsa ⎤ ⎡va ⎤ ⎡ia ⎤
⎢v ⎥ = ⎢v ⎥ + L ⋅ d ⎢i ⎥
(4)
⎢ sb ⎥ ⎢ b ⎥ s
dt ⎢ ⎥
b

⎣⎢vsc ⎦⎥ ⎣⎢vc ⎦⎥ ⎣⎢ic ⎦⎥

Through the dq axis decomposition (1), equation (4) in dq

frame is in the form:
Line Inverter
Impedance Filter Inverter ⎡vsd ⎤ ⎡vd ⎤ ⎡ Ls p − ω PLL Ls ⎤ ⎡id ⎤
⎢v ⎥ = ⎢v ⎥ + ⎢ ⋅⎢ ⎥ (5)
⎣ sq ⎦ ⎣ q ⎦ ⎣+ ω PLL Ls Ls p ⎥⎦ ⎣iq ⎦
DC
Source
The constant current inverter control is composed only by a
Grid Local Gate Signals current regulator. This regulator is a PI controller. The current
Load Inverter regulator acts canceling the error between the current input
Controller references idref and iqref and the current injection of the IBDG id
Fig. 3. Three-phase grid-connected VSI – switched model. and iq. The output of current regulator is the compensation of
the voltage drop across the harmonic filter, so the voltage of
the controlled source is the terminal voltage imposed by the
system plus the voltage drops in the filter inductance Ls. The
control diagram block is shown in Fig. 6. Equations (6)-(9)
describes the dynamics of the control.

Electric Circuits Linear Control ud = Ls ⋅ p ⋅ id (6)

u q = Ls ⋅ p ⋅ iq (7)
Controlled Voltage
⎛ k ⎞
Source
u d = ⎜ k pi + ii ⎟ ⋅ (idref − id ) (8)
⎝ s ⎠
dq
⎛ kii ⎞
Inverter u q = ⎜ k pi + ⎟ ⋅ (iqref − iq ) (9)
Control ⎝ s ⎠
abc

Grid/Load

abc

PLL dq

Fig. 4. Three-phase VSI - average model.

 GEVS

Current Regulator Current Regulator

Fig. 6. Diagram block of the current controller of VSI. Fig. 8. GEVS extensions to constant current-controlled three-phase inverter
systems.

Positive Feedback Anti-islanding Scheme Models

Load and Network Model
The positive feedback schemes previously introduced will be
Consider the system showed in Fig. 9, an IBDG grid-connected
now modeled to be included in the IBDG model under
with a RLC parallel local load.
development.
The KVL applied to the mesh formed by the system bus, the
GEVS. A small perturbation idp is inserted into direct axis
distribution line and the IBDG is:
current reference idref in order to establish the voltage positive
feedback described in Fig. 2. This perturbation on direct axis
current reference is driven by the direct axis voltage deviation
⎡va ⎤ ⎡iNa ⎤ ⎡iNa ⎤ ⎡ea ⎤
Δvd as shown in (10). ⎢ v ⎥ = R ⋅ ⎢i ⎥ + L ⋅ d ⎢i ⎥ + ⎢ e ⎥
⎢ b⎥ ⎢ Nb ⎥ (11)
dt ⎢ ⎥ ⎢ ⎥
Nb b

idp = K v ⋅ H BPF (s ) ⋅ vd (10) ⎣⎢ vc ⎦⎥ ⎣⎢iNc ⎦⎥ ⎣⎢iNc ⎦⎥ ⎣⎢ ec ⎦⎥

Figure 7 shows the GEVS diagram block and Fig. 8 shows the
modification into inverter control because of GEVS inclusion.

Band-Pass
Filter Gain
Fig. 9. Three-phase diagram of the IBDG connected to the system through an
R+jωL distribution line feeding a RLLLCL parallel local load.

Or in dq coordinates:
BPF
Fig. 7. GEVS diagram block.

⎡vd ⎤ ⎡ R + Lp − ωPLL L⎤ ⎡iNd ⎤ ⎡ed ⎤

⎢v ⎥ = ⎢ ⎥⋅⎢ ⎥ + ⎢ ⎥ (12)
⎣ q ⎦ ⎣+ ωPLL L R + Lp ⎦ ⎣iNq ⎦ ⎣eq ⎦
The Kirchhoff’s Current Law (KCL) applied to the PCC:
 Model Linearization
⎡ia ⎤ ⎡iRa ⎤ ⎡iLa ⎤ ⎡iCa ⎤ ⎡iNa ⎤
⎢i ⎥ = ⎢i ⎥ + ⎢i ⎥ + ⎢i ⎥ + ⎢i ⎥ The average model of the IBDG constant current controlled
(13)
⎢ b ⎥ ⎢ Rb ⎥ ⎢ Lb ⎥ ⎢ Cb ⎥ ⎢ Nb ⎥ with GEVS anti-islanding control represented by the
⎢⎣ic ⎥⎦ ⎢⎣iRc ⎥⎦ ⎢⎣iLc ⎥⎦ ⎢⎣iCc ⎥⎦ ⎢⎣iNc ⎥⎦
previously 17 equations (2)-(3), (6)-(9), (10), (12),(14) and
(18)-(20) can be linearized around x0 as follow:
Or in dq coordinates:
y = f ( x, t , u ) (23)
⎡id ⎤ ⎡iRd ⎤ ⎡iLd ⎤ ⎡iCd ⎤ ⎡iNd ⎤
⎢i ⎥ = ⎢i ⎥ + ⎢i ⎥ + ⎢i ⎥ + ⎢i ⎥ (14) y (t ) = y0 (t ) + Δy(t ) (24)
⎣ q ⎦ ⎣ Rq ⎦ ⎣ Lq ⎦ ⎣ Cq ⎦ ⎣ Nq ⎦ y0 (t ) = f ( x0 , t , u0 ) (25)
∂f ( x, t , u ) ∂f ( x, t , u )
Finally, the RL, LL and CL dynamic equations: Δy (t ) = ⋅ Δx + ⋅ Δu (26)
∂x x 0 ,u 0 ∂u x 0 ,u 0

⎡iRa ⎤ ⎡va ⎤
⎢i ⎥ = 1 ⋅ ⎢v ⎥ The steady-state operation point x0 can be obtained through a
(15)
⎢ Rb ⎥ R ⎢ b ⎥ steady-state analysis like a load flow.
⎣⎢iRc ⎦⎥ ⎢⎣vc ⎦⎥
L

⎡ a⎤
v ⎡iLa ⎤ The linearized equations are then:
⎢v ⎥ = L ⋅ d ⎢i ⎥
(16)
⎢ b⎥ L
dt ⎢ ⎥
Lb

⎢⎣vc ⎥⎦ ⎢⎣iLc ⎥⎦ Δvd = RΔiNd + LpΔiNd − ω0 LΔiNq − LiNq0 ΔωPLL − E sin(δ 0 )Δδ PLL (27)
Δvq = RΔiNq + LpΔiNq + ω0 LΔiNd + LiNd 0 ΔωPLL − E cos(δ 0 )Δδ PLL (28)
⎡iCa ⎤ ⎡v a ⎤
⎢i ⎥ = C ⋅ d ⎢v ⎥ Δid = ΔiRd + ΔiLd + ΔiCd + ΔiNd (29)
(17)
⎢ Cb ⎥ L
dt ⎢ ⎥
b
Δiq = ΔiRq + ΔiLq + ΔiCq + ΔiNq (30)
⎣⎢iCc ⎦⎥ ⎣⎢vc ⎦⎥
ΔiRd = (1 / RL )Δvd (31)
Or in dq coordinates: ΔiRq = (1 / RL )Δvq (32)
Δvd = LL pΔiLd − ω0 LL ΔiLq − LL iLq 0 ΔωPLL (33)
⎡iRd ⎤ 1 ⎡vd ⎤ Δvq = LL pΔiLq + ω0 LL ΔiLd + LLiLd 0 Δω PLL (34)
⎢i ⎥ = ⋅⎢ ⎥ (18)
⎣ Rq ⎦ RL ⎣vq ⎦ ΔiCd = C L pΔvd − ω0CL Δvq − CL vq 0 ΔωPLL (35)
⎡vd ⎤ ⎡ LL p − ω PLL LL ⎤ ⎡iLd ⎤ ΔiCq = CL pΔvq + ω0CL Δvd + CL vd 0 ΔωPLL (36)
⎢v ⎥ = ⎢ ⋅⎢ ⎥ (19)
⎣ q ⎦ ⎣+ ω PLL LL LL p ⎥⎦ ⎣iLq ⎦ Δud = Ls pΔid (37)
⎡iCd ⎤ ⎡ CL p − ω PLLCL ⎤ ⎡vd ⎤ Δu q = Ls pΔiq (38)
⎢i ⎥ = ⎢ ⋅⎢ ⎥ (20)
⎣ Cq ⎦ ⎣+ ω PLLC CL p ⎥⎦ ⎣vq ⎦ pΔu d = k pi p (Δidref + Δidp − Δid ) + kii (Δidref + Δidp − Δid ) (39)
pΔuq = k pi p (Δiqref − Δiq ) + kii (Δiqref − Δiq ) (40)
The (vd,vq) versus (id, iq) DG terminal characteristics is given
pΔω PLL = k pPLL pΔvq + kiPLL Δvq (41)
by equation (5). Assuming the equivalent distribution system
bus as a perfectly balanced three-phase voltage source, stiff, pΔδ PLL = Δω PLL (42)
with constant angular frequency ω0 and magnitude E, then: Δidp = K v Δvd (43)

⎡ Esin(ω t ) ⎤ The set (27)–(43) can be rearranged to be expressed in a matrix

⎢ 0
⎥ form:
⎡ea ⎤ ⎢ ⎛ 2 ⎞⎥
⎢e ⎥ = ⎢ Esin⎜ ω0t − π ⎟ ⎥
⎝ 3 ⎠⎥ (21)
⎢ b⎥ ⎢ E p Δ x = AΔ x + B Δ u (44)
⎣⎢ec ⎦⎥ ⎢ 2 ⎥
⎢ Esin⎛⎜ ω0t + π ⎞⎟⎥
⎣⎢ ⎝ 3 ⎠⎦⎥ Δx is the state vector and Δu, the input vector to the constant
current injection control, is defined as:
Or in dq coordinates:
Δu = [Δidref Δiqref ]T (45)
⎡ed ⎤ ⎡ E cos(δ PLL ) ⎤
⎢e ⎥ = ⎢ ⎥ (22)
⎣ q ⎦ ⎣ − Esin(δ PLL )⎦ As the model has input derivative, these inputs must be added
to the state vector ∆x, since in state space representation only
state variables can be derivate.

Δx = [ Δvd Δvq Δid Δiq Δud Δuq ΔiNd ΔiNq ΔiRd

(46)
ΔiRq ΔiLd ΔiLq ΔiCd ΔiCq Δω PLL Δδ PLL Δidp Δidref Δiqref ]T
The addition of two extra state variables requires the inclusion id Step Response
of two other equations in the set (27)–(43). These equations are
ref
0.02
the equivalence between the input idref and iqref and the state Average Model
0.018
variables idref and iqref. Small−Signal Model

0.016
Model Validation 0.014

The linear model developed in the previous section is 0.012

Δid (p.u.)
confronted against the average model of the inverter so we can 0.01
conclude about your validation.
0.008

Time Domain Dynamic Response for a Reference Step 0.006

0.004
The single DG system shown in Fig. 9 average model were
implemented in SimPowerSystems™/Matlab®. Equation (44), 0.002
which contains both Differential and Algebraic Equations
0
(DAE), was integrated in time domain using Matlab® ode15s, 2 2.02 2.04 2.06 2.08 2.1
Time (s)
a variable order solver based on the numerical differentiation
Fig. 11 Dynamic response of direct axis current id.
formulas (NDF). Because of the model’s algebraic equations,
matrix E is singular and ode15s can solve DAE with singular id Step Response
mass matrix. ref
0.02
Average Model
The parameters used in the single DG model are listed in Table 0.018 Small−Signal Model
1 and the response to a step change of 10% in idref is shown in 0.016
Fig. 10–Fig. 13.
0.014

0.012
Table 1. Parameters used in single DG simulations.
ΔP (p.u.)

Parameter Value Parameter Value 0.01

Vbase (rms, ph-ph) 208 V kpPLL 50
0.008
Sbase (3-ph) 100 kW kiPLL 500
R 0.2 Ω idref 0.1 pu 0.006
X 0.3 Ω iqref 0.0 pu
Ls 1.0 mH kvGEVS 0.20 0.004
kpi 0.5 RL 4.32 Ω
0.002
kii 500 LL 6.37 mH
CL 1.11 mF 0
2 2.02 2.04 2.06 2.08 2.1
Time (s)
Fig. 12 Dynamic response of active power P.
i Step Response
d
ref
id Step Response
0.015 ref
Average Model 1.2
Small−Signal Model Average Model
0.01
1 Small−Signal Model

0.005
0.8
Δvd (p.u.)

0
0.6
Δω (rad/s)

−0.005 0.4

−0.01 0.2

0
−0.015

−0.2
−0.02
2 2.02 2.04 2.06 2.08 2.1
Time (s) −0.4
Fig. 10. Dynamic response of direct axis voltage vd. 2 2.02 2.04 2.06 2.08 2.1
Time (s)
Fig. 13 Dynamic response of angular frequency ω.
One can assess the stability of the IBDG grid-connected Proc. 2007 IEEE Power Engineering Society General Meeting, pp.1-4,
24-28 Jun 2007.
operation because of the positive feedback control through the
[13] E. Virnik, “Stability Analysis of Positive Descriptor Systems” Linear
eigenanalysis of matrix pencil A,E in equation (44). The Algebra and its Applications, vol.429, no.10, pp.2640-2659, Nov. 2008.
system is asymptotically stable if all its finite generalized
eigenvalues have negative real part [13]. The generalized Acknowledgments
eigenvalues λ of the matrix pencil A, E are the roots of the
characteristic equation: The authors would like to thanks São Paulo State Research
Foundation (FAPESP) and Coordination for the Improvement
det (A − λE) = 0 (47)
of Higher Level Personnel (CAPES) for the support of this
work.
Conclusions
Tiago R. Ricciardi is M.Sc. graduate student with the Department of Electrical
Energy Systems of University of Campinas, Campinas, SP, Brazil. His research
The presented small-signal model is useful to analyze the
interests are distributed generation and power systems stability and control. (e-
impact of voltage positive-feedback anti-islanding schemes on mail: [email protected]).
grid-connected IBDG systems stability. The linearized set of Diogo Salles is Ph.D. graduate student with the Department of Electrical
equations can well represent the system against small Energy Systems of University of Campinas, Campinas, SP, Brazil. His research
perturbations, with the advantages of the state-space interests are distributed generation, distribution systems and power systems
protection, stability and control. (e-mail: [email protected]).
representation. This state-space representation allows the use of
a whole set of linear control techniques to direct stability Walmir Freitas is Associate Professor with the Department of Electrical
assessment, without the need of slow time domain simulations. Energy Systems of University of Campinas, Campinas, SP, Brazil. His research
interests include distributed generation and distribution systems. (e-mail:
[email protected]).
Through the modal analysis of the DAE representation of the
system one can conduct an extensive parametric sensitivity Xiaoyu Wang is Assistant Researcher with the Department of Electrical
study in order to assess the limit penetration of multiple IBDGs Engineering at Tsinghua University, Beijing, China. His research interests
in distribution systems. include distributed generation and power system stability. (e-mail:
[email protected]).

References
[1] N. Jenkins, R. Allan, P. Crossley, D. Kirschen e G. Strbac, “Embedded
Generation”, London: The Institution of Electrical Engineers (IEE), 2000,
292 p.
[2] A. M. Borbely e J. F. Kreider, "Distributed Generation: The Power
Paradigm for the New Millennium", Boca Raton, FL: CRC Press, 2001,
416 p.
[3] IEEE Standard for Interconnecting Distributed Resources with Electric
Power Systems, IEEE Standard 1547-2003, New York, Jun 2003.
[4] IEEE Recommended Practice for Utility Interface of Photovoltaic (PV)
Systems, IEEE Standard 929-2000, New York, Jan 2000.
[5] W. Bower e M. Ropp, “Evaluation of Islanding Detection Methods for
Utility-Interactive Inverters in Photovoltaic Systems” Sandia National
Laboratories, Albuquerque, NM, Tech. Rep. SAND 2002-3591, Nov.
2002.
[6] J. Stevens, R. Bonn, J. Ginn, S. Gonzalez e G. Kern, “Development and
Testing of an Approach to Anti-Islanding in Utility-Interconnected
Photovoltaic Systems” Sandia National Laboratories, Albuquerque, NM,
Tech. Rep. SAND 2000-1939, Aug. 2000.
[7] Z. Ye, R. Walling, L. Garces, R. Zhou, L. Li e T. Wang, "Study and
Development of Anti-Islanding Control for Grid-Connected Inverters"
National Renewable Energy Laboratory (NREL), Golden, CO, Tech.
Rep. NREL/SR-560-36243, May. 2004.
[8] X. Wang, “Investigation of Positive Feedback Anti-Islanding Scheme for
Inverter-Based Distributed Generation” Ph.D. thesis, Dept. Elect. and
Comp. Eng., Univ. Alberta, Edmonton, 2008.
[9] W. Freitas, “Métodos para Análise de Proteção Anti-ilhamento de
Geradores Distribuídos” Thesis, School of Elect. and Comp. Eng., Univ.
of Campinas, 2007.
[10] X. Wang, W. Freitas, W. Xu e V. Dinavahi, "Impact of DG Interface
Controls on the Sandia Frequency Shift Antiislanding Method" IEEE
Transactions on Energy Conversion, vol.22, no.3, pp.792-794, Sep. 2007.
[11] X. Wang e W. Freitas, "Impact of Positive-Feedback Anti-Islanding
Methods on Small-Signal Stability of Inverter-Based Distributed
Generation" IEEE Transactions on Energy Conversion, vol.23, no.3,
pp.923-931, Sep. 2008.
[12] X. Wang, W. Freitas, W. Xu e V. Dinavahi, "Impact of Interface Controls
on the Steady-State Stability of Inverter-Based Distributed Generators" in