0% found this document useful (0 votes)
24 views11 pages

Robust Mathcal H Infty State Feedback Controllers Based On Linear Matrix Inequalities Applied To Grid-Connected Converters

Uploaded by

Sivadharshini A
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
24 views11 pages

Robust Mathcal H Infty State Feedback Controllers Based On Linear Matrix Inequalities Applied To Grid-Connected Converters

Uploaded by

Sivadharshini A
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 11

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 66, NO.

8, AUGUST 2019 6021

Robust H∞ State Feedback Controllers Based


on Linear Matrix Inequalities Applied
to Grid-Connected Converters
Gustavo Guilherme Koch , Student Member, IEEE, Luiz Antonio Maccari Jr.,
Ricardo C. L. F. Oliveira , and Vinı́cius Foletto Montagner

Abstract—This paper provides robust H∞ state feedback described in [5], which specifies acceptable limits in terms of
controllers suitable for implementation in three-phase total harmonic distortion (THD), responses to abnormal grid
grid-connected converters. This control strategy is known conditions, and anti-islanding strategies [6]–[10].
to provide optimal rejection of disturbances, but usually
leads to high control gains, that may be difficult to be imple- The pulsewidth modulation (PWM) voltage produced by in-
mented in practice. To mitigate this problem, a linear matrix verters needs to be filtered to prevent grid currents with pro-
inequality condition based on slack variables is proposed, hibitive harmonic content. One suitable solution is the low-pass
which allows to impose bounds on the control gains in a less LCL filter, due to its 60 dB per decade attenuation, with relatively
conservative way than conventional quadratic stability. The
small values of inductance and capacitances. Nevertheless, these
performance is proven to be superior to similar H∞ state
feedback controllers in the literature, providing an upper filters have a resonance peak that needs to be damped in order
bound for the converter output admittance and experimen- to prevent poor performances or instability. Passive damping is
tal grid currents complying with the IEEE Standard 1547. an alternative, reducing the resonance peak by inserting resis-
Index Terms—Grid-connected converters, H∞ control, tive elements in the LCL filter [11], even taking into account
LCL filters, linear matrix inequalities (LMIs). optimization constraints [12], providing stable behavior, but at
the price of decreasing efficiency. To overcome this problem,
I. INTRODUCTION active damping has been used [13]–[21]. Among the most used
HE increasing demand for electrical energy and environ- current controller techniques [13], one has the proportional-
T mental issues motivates the expansion in the use of renew-
able energy sources, as wind- and photovoltaic-based generation
integral, which deals with the problem for constant references,
in asynchronous axes [14], [22], and the proportional-resonant,
[1], [2]. Power electronics is capable of providing an interface which deals with the problem for sinusoidal references, in syn-
between renewable energy sources and the grid, allowing the chronous axes [15]–[18]. In [19]–[21], frequency domain based
control of the power flow and the control of electrical variables approaches are used to design the controller relying on the feed-
as currents, voltages, and frequencies [3], [4]. In this scenario, back of the current in the filter capacitor. However, a common
grid-connected converters have great importance. These systems point in the above cited papers is that the robustness to grid pa-
include inverters, for dc–ac conversion, and filters, for reduc- rameter uncertainty is ensured a posteriori, testing the closed-
tion of harmonics in the grid-side variables. For the operation loop system under several grid impedance situations, but varia-
connected to the grid, these converters have to ensure that the tions in the grid impedance are usually not taken into account.
electrical variables comply with stringent standards, as the ones In order to provide controllers, which are robust to paramet-
ric uncertainties and variations, linear matrix inequalities (LMIs,
[23]) are a useful design tool, mainly when the synthesis con-
Manuscript received February 19, 2018; revised May 18, 2018 and
July 4, 2018; accepted September 2, 2018. Date of publication Septem- ditions are based on parameter-dependent Lyapunov functions
ber 24, 2018; date of current version March 29, 2019. This work was [24]–[26], which emerged as an alternative to reduce the conser-
supported in part by the Brazilian Coordination for the Improvement of vativeness associated with the first methods based on the classic
Higher Education Personnel (CAPES), in part by the National Council
of Technological and Scientific Development (CNPq, Research Project notion of quadratic stability [23]. LMI-based conditions allow
306197/2015-4), and in part by the National Institute of Science and us to deal with the synthesis of controllers to grid-connected
Technology – Distributed Generation Power Systems (INCT–GD). (Cor- converters, including performance constraints in terms of pole
responding author: Gustavo Guilherme Koch.)
G. G. Koch and V. F. Montagner are with the Federal University of Santa location, optimization of quadratic cost functions, etc. [27]–
Maria, Santa Maria 97105-900, Brazil (e-mail:, gustavoguilhermekoch@ [29]. In the scenario of optimal control, the H∞ state feedback
gmail.com; [email protected]). strategy is known to provide optimal rejection of disturbances
L. A. Maccari Jr. is with the Federal University of Santa Catarina,
Florianopolis 88040-900, Brazil (e-mail:, [email protected]). [30]. One drawback of this technique is that the control signal is
R. C. L. F. Oliveira is with the University of Campinas, Campinas often difficult to be implemented in practice by means of PWM
13083-970, Brazil (e-mail:, [email protected]). inverters, due to high control gains, which lead to amplitude
Color versions of one or more of the figures in this paper are available
online at http://ieeexplore.ieee.org. saturation or demand components of prohibitively high frequen-
Digital Object Identifier 10.1109/TIE.2018.2870406 cies. One can relax optimality, in order to get suboptimal H∞
0278-0046 © 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications standards/publications/rights/index.html for more information.

Authorized licensed use limited to: Seoul National Univ of Science & Tech (SNUT). Downloaded on July 04,2023 at 07:54:21 UTC from IEEE Xplore. Restrictions apply.
6022 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 66, NO. 8, AUGUST 2019

Fig. 1. (a) Three-phase inverter connected to the grid by means of LCL filter. (b) Control implementation in αβ coordinates.

controllers, for instance, by adding constraints on the size of the predominantly inductive, with inductance Lg 2 uncertain, be-
control gains, which, although do not limit directly the control longing to an interval whose limits are known. In a general way,
signal amplitude, can help to mitigate the problem of actuator the problem to be solved in this paper is to obtain a control law
saturation. It is worth to mention that few results are available to synthesize the inverter output voltages in Fig. 1(a) such that
in the literature showing cases of H∞ state feedback controllers the grid-injected currents follow sinusoidal references, with low
applied to grid-connected converters subject to uncertain pa- harmonic distortion, under grid inductances (Lg 2 ) not precisely
rameters, with high performance confirmed by experimental known (uncertain parameters) and grid voltages (vg ) possibly
results [31]–[33]. In [31], one has a successful application of affected by harmonics (disturbances).
H∞ state feedback controllers for uninterruptible power sup- It is assumed that the inductor currents and capacitor voltages
plies, with experimental voltage with THD in accordance to a of the LCL filter are available for feedback. The current control
pertinent standard. Specifically for grid-connected converters, is implemented in αβ coordinates, in a state feedback based
one has the results in [32] and [33] for single- and three-phase strategy, for each coordinate α or β, as shown in Fig. 1(b).
inverters, respectively, with experimental results given only for In this strategy, Kx represents the gains of the plant states,
short circuit tests. One can notice the low number of works with Kφ represents the gain of the delayed control, and Kξ are the
high performance confirmed experimentally with this control gains of the internal states of resonant controllers [27], all of
technique. them to be provided by the robust control strategy, given in
This paper provides as main contribution the design and ex- Section III.
perimental validation of a H∞ state feedback controller for From Fig. 1(a), considering the average model of the PWM
grid-connected converters based on LMIs formulated in terms inverter and that all inductors and capacitors operate linearly,
of parameter-dependent Lyapunov functions and slack variables, then, from Kirchhoff’s laws, one can write the state space
which allows us to impose bounds on the control gains in a less model [1], [4]
conservative way than the quadratic stability based results. The
controller is robust to grid parameter uncertainties and varia- ẋ = Aabc xabc + Bu abc uabc + Bdabc vgabc (1)
tions, with experimental results complying with the IEEE Stan-
with matrices Aabc , Buabc , and Bdabc given by
dard 1547. The rest of this paper is organized as follows. Sec-
⎡ ⎤
tion II presents a polytopic model of the system, including delay 03×3 Ap1 03×3
and resonant controllers with arbitrary dimension, and the for- ⎢ ⎥
Aabc = ⎣ Ap2 03×3 −Ap2 ⎦
mulation of the control problem in terms of the H∞ norm, which
gives a limit for the output admittance of the converter for the en- 03×3 Ap3 03×3
tire frequency range. Section III presents four LMI conditions to ⎡ ⎤ ⎡ ⎤
−Ap1 03×3
synthesize H∞ state feedback controllers, with or without slack ⎢ ⎥ ⎢ ⎥
variables and limitation of the control gains. Section IV presents Buabc = ⎣ 03×3 ⎦ ; Bdabc = ⎣ 03×3 ⎦ (2)
comparisons among these design conditions, showing that the 03×3 −Ap3
LMIs with slack variables and with limitation of the control
gains provide the better results. Section VI provides practical and matrices Ap1 , Ap2 , and Ap3 written as
results in a three-phase experimental prototype, showing grid ⎡ ⎤
2 1 1
currents with good correspondence with simulations and with −
⎢ 3Lc 3L 3L ⎥
harmonic content complying with the IEEE Standard 1547. ⎢ c c ⎥
⎢ 1 2 1 ⎥
⎢ ⎥
Ap1 = ⎢ − ⎥;
II. PROBLEM FORMULATION ⎢ 3Lc 3Lc 3Lc ⎥
⎢ ⎥
⎣ 1 1 2 ⎦
A three-phase inverter connected to the grid by means of −
an LCL filter is shown in Fig. 1(a). The grid is assumed 3Lc 3Lc 3Lc

Authorized licensed use limited to: Seoul National Univ of Science & Tech (SNUT). Downloaded on July 04,2023 at 07:54:21 UTC from IEEE Xplore. Restrictions apply.
KOCH et al.: ROBUST H∞ STATE FEEDBACK CONTROLLERS BASED ON LMIs APPLIED TO GRID-CONNECTED INVERTERS 6023

⎡ 1 ⎤ ⎡ ⎤
0 0 0 ⎡ ⎤
⎢ Cf ⎥ ⎢ ⎥ icα
⎢ ⎥ ⎢ ⎥
⎢ 1 ⎥ ⎢ 0 ⎥ ⎢ ⎥
⎢ ⎥ Bd = ⎢ ⎥ , xα = ⎣ vcα ⎦ . (9)
Ap2 =⎢ 0 0 ⎥; ⎢ ⎥
⎢ Cf ⎥ ⎣ 1 ⎦
⎢ ⎥ −
ig α
⎣ 1 ⎦ Lg
0 0
Cf A model identical to (8) and (9) is valid for axis β.
⎡2 1 1 ⎤ The problem now is to ensure that the grid currents ig (in axes
− −
⎢ 3Lg 3Lg 3Lg ⎥ α and β) track sinusoidal references and, for that, the control
⎢ ⎥
⎢ 1 2 1 ⎥ law u (in axes α and β) must be properly synthesized.
⎢ ⎥
Ap3 = ⎢− − ⎥ (3) For the application of a digital control law, consider the dis-
⎢ 3Lg 3Lg 3Lg ⎥
⎢ ⎥ cretization of the plant with a sufficiently small sampling period
⎣ 1 1 2 ⎦
− − Ts , and also the inclusion of an additional state representing the
3Lg 3Lg 3Lg transport delay in the application of the control signal1
being x(k + 1) = A(θ)x(k) + Bu (θ)φ(k) + Bd (θ)vg (k)
Lg = Lg 1 + Lg 2 (4) φ(k + 1) = u(k) (10)
the uncertain parameter. where φ represents the delay.
The state vector xabc , the control vector uabc , and the distur- To ensure tracking of sinusoidal references and rejection of
bance vector vgabc are given by harmonic disturbances, n resonant controllers are included in

xabc = [ ica icb icc vca vcb vcc ig a ig b ig c ] the model, leading to the representation

uabc = [ ua ub uc ]

ξ(k + 1) = Rξ(k) + T (ir ef (k) − y(k))
 y(k) = Cx(k) = [001]x(k) = ig (k) (11)
vgabc = [ vg a vg b vg c ] . (5)
Assuming a balanced three-phase system, the previous model where iref is the reference for the grid currents, and ξ, R, and T
can be rewritten from abc to αβ0 coordinates by means of the are, respectively, the state vector and the matrices of the resonant
transformation [34] controller (for more details in model (11), see [27, Section II]).
⎡ ⎤ From the internal model principle, one has that if the closed-
1 1
⎢ 1 − − loop system is asymptotically stable and if the controlled plant
⎢ 2 2 ⎥
√ ⎥
contains all marginally stable modes of the reference signal
⎢ √ ⎥
⎢ 3 3 ⎥. and disturbance, the tracking of a sinusoidal reference and the
Tα β = ⎢ 0 − ⎥ (6)
⎢ 2 2 ⎥ rejection of harmonic disturbances is ensured [35], [36].
⎢ ⎥
⎣1 1 1 ⎦
Notice that (10) can be rewritten as
⎡ ⎤ ⎡ ⎤
2 2 2 x(k + 1) A(θ) Bu (θ) 03×2n
⎢ ⎥ ⎢ ⎥
Since there is no path for the current of axis “0,” the system ⎣ φ(k + 1) ⎦ = ⎣ 01×3 0 01×2n ⎦
can be represented by
     ξ(k + 1) −T C 02n ×1 R
ẋα A 03×3 xα Bu 03×1 uα ⎡ ⎤⎡ ⎤ ⎡ ⎤
= + x(k) 03×1 Bd (θ)
ẋβ 03×3 A xβ 03×1 Bu uβ ⎢ ⎥⎢ ⎥ ⎢ ⎥
× ⎣ φ(k) ⎦ ⎣ 1 ⎦ u(k) + ⎣ 0 ⎦ vg (k)
 
Bd 03×1 vg α ξ(k) 02n ×1 02n ×1
+ (7)
03×1 Bd vg β ⎡ ⎤
03×1
⎢ ⎥
Notice that (7) represents two single-phase decoupled systems. + ⎣ 0 ⎦ iref (k) (12)
For instance, for axis α, this model is given by
T
ẋα = Axα + Bu uα + Bd vg α (8)
or, in a more compact form, as
with
⎡ ⎤ ρ(k + 1) = G(θ)ρ(k) + Hu (θ)u(k) + Hd (θ)vg (k)
1 ⎡ ⎤
0 − 0 1
⎢ Lc1 ⎥ + Hr ef (θ)ir ef )
⎢ ⎥ ⎢ Lc1 ⎥
⎢ 1 1 ⎥ ⎢ ⎥
A=⎢ 0 − ⎥ , Bu = ⎢ 0 ⎥, y(k) = Caug ρ(k) , Caug = C 01×(2n +1) (13)
⎢ Cf ⎥
Cf ⎥ ⎢ ⎥
⎢ ⎣ ⎦
⎣ 1 ⎦
0 0 0 1 From this point on, for simplicity, the subscripts α and β are suppressed in
Lg the variables, which depend on (k).

Authorized licensed use limited to: Seoul National Univ of Science & Tech (SNUT). Downloaded on July 04,2023 at 07:54:21 UTC from IEEE Xplore. Restrictions apply.
6024 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 66, NO. 8, AUGUST 2019

where III. H∞ LMI-BASED DESIGN CONDITIONS


2 Problem 1 can be solved efficiently by means of LMIs that si-
(G, Hu , Hd , Hr ef )(θ) = θj (Gj , Hu j , Hdj , Hrefj ) (14) multaneously guarantee limitation on the H∞ norm (property 1)
j =1 and closed-loop pole location (property 2).
2 However, one usual drawback for the practical application of
θj = 1, θj ≥ 0, j = 1, 2. (15) optimal H∞ controllers is in the high control gains. To overcome
j =1 this difficulty, the design conditions of these controllers can be
modified, as indicated in this section.
Notice that (13)–(15) is a polytopic representation, where Specifically, in the sequence, one describes in terms of feasi-
θ is used to represent Lg . In (14), each matrix is written as bility problem of LMIs, four different conditions for computa-
a convex combination of the vertices, for instance, G(θ) = tion of gain K (named FEAS1, FEAS2, FEAS3, and FEAS4),
θ1 G1 + θ2 G2 , θ1 ≥ 0, θ2 ≥ 0 and θ1 + θ2 = 1 [23]. This repre- based on H∞ control with pole location: with and without slack
sentation is suitable to design robust state feedback controllers, variables, with and without limitations on control gains.2
given by FEAS 1: Given γ > 0 and 0 < r ≤ 1 if there exist a sym-
⎡ ⎤ metric positive definite matrix W and a matrix Z such that
x(k) ⎡ ⎤
⎢ ⎥ Gi Hu i
u(k) = Kx Kφ Kξ ⎣ φ(k) ⎦ = Kρ(k).
⎢W W+ Z 0
(16) Hdi ⎥
⎢ r r ⎥
ξ(k) ⎢  ⎥
⎢ W WCaug 0 ⎥ > 0, i = 1, 2; (20)
⎢ ⎥
Considering as the input vg , since the grid can have harmonics ⎢ 0 ⎥
⎣ I ⎦
to be rejected, and taking into account (13) and (16), the closed-
loop system can be written as γ2 I
then
ρ(k + 1) = Gcl (θ)ρ(k) + Hd (θ)vg (k)
K = ZW −1 (21)
y(k) = Caug ρ(k) = ig (k) (17)
is a robust gain for state feedback control law (16) that ensures
with Gcl (θ) = G(θ) + Hu (θ)K. (18) and (19).
The specific problem of control design to be solved in this FEAS 2: Given γ > 0 and 0 < r ≤ 1, and kw > 0 and kz >
work is now described. 0, if there exist a symmetric positive definite matrix W and a
Problem 1: Determine K, such that matrix Z such that
1)
LMI (20);
y(k)2  
≤ γ, w(k) ∈ 2 (18) kw I −kz I (22)
w(k)2 > 0; <0
I W Z −I
2)
then (21) is a robust gain for state feedback control law (16) that
max |λ (Gcl (θ))| < r, 0 < r ≤ 1 (19) ensures (18) and (19).
FEAS 3: Given γ > 0 e 0 < r ≤ 1, if there exist symmetric
where w(k) ∈ 2 is any disturbance signal with bounded energy positive definite matrices Si and matrices F and G such that
[represented by vg (k) in (17)], and λ represents the eigenvalues ⎡  ⎤

of matrix Gcl (θ) for all θ satisfying (15).   Gj  Hu j  
⎢ G + G − S j 0 G + F G Caug ⎥
Taking the supremum value of y(k)2 /w(k)2 , one has ⎢ r r ⎥
that the smallest γ fulfilling (18) is the L2 gain of the sys- ⎢ ⎥ i = 1, 2;
⎢ I Hdj  0 ⎥ > 0;
⎢ ⎥ j = 1, 2
tem (17) which, in the time-invariant case, can also be in- ⎢ Si 0 ⎥
terpreted as the H∞ norm of the transfer function from the ⎣ ⎦
input w to the output y. In the particular SISO system inves- γ2 I
tigated here, property 1) also ensures that the frequency re- (23)
sponse from input vg to output ig always has magnitudes lower then
than γ (peak of the Bode diagram) [23]. Therefore, property
1) guarantees an upper bound for the converter output admit- K = FG −1 (24)
tance seen by source vg . Property 2) ensures that all closed-
is a robust gain for state feedback control law (16) that ensures
loop eigenvalues have modules lower than r, for all θ sat-
(18) and (19).
isfying (15), which guarantees that the closed-loop transient
FEAS 4: Given γ > 0 and 0 < r ≤ 1, μ0 > 0, and δ > 0, if
response modes are bounded by rk . Therefore, property 2)
there exist symmetric positive definite matrices Si and matrices
ensures robust stability against grid uncertain parameter and
also a limit for the settling time of the closed-loop transient
responses. 2 The proof for the conditions in this section is given in the Appendix.

Authorized licensed use limited to: Seoul National Univ of Science & Tech (SNUT). Downloaded on July 04,2023 at 07:54:21 UTC from IEEE Xplore. Restrictions apply.
KOCH et al.: ROBUST H∞ STATE FEEDBACK CONTROLLERS BASED ON LMIs APPLIED TO GRID-CONNECTED INVERTERS 6025

G and F such that TABLE I


SYSTEM PARAMETERS
LMI (23);

G + G  − μ0 I (25)
>0
F δI
then (24) is a robust gain for state feedback control law (16) that
ensures (18) and (19).
Notice that FEAS1 and FEAS2 do not use slack variables,
while FEAS3 and FEAS4 use slack variables. Notice also that
FEAS1 and FEAS3 do not provide limitation of the control gain,
while FEAS2 and FEAS4 provide this limitation.
It is important to note that each of the above conditions can
be used to get optimal H∞ controllers by solving the following
convex optimization problem: input source of 400 V, a three-phase LCL filter connected, at
OPT 1: γ ∗ = min γ s.t. (20) or (23). each phase, with a sinusoidal voltage source in series with an
However, these optimal H∞ controllers generally lead to high inductor, representing the grid. The filter state variables are mea-
control gains, which may not be viable in practice for this ap- sured and used as inputs in a dynamic link library (DLL) block
plication, as will be shown in Section IV. programmed in C++ language, which includes also the control
More interesting optimizations for H∞ controllers for this gains computed in MATLAB, and the equations of resonant
application are given in the problems: controllers, here used at the frequencies of the fundamental,
OPT 2: min kz + kw s.t. (20) and (22) third, fifth, and seventh harmonics. The control algorithm com-
OPT 3: min δ − μ0 s.t. (23) and (25) putes and limits the control signal, which is then converted in
allowing us to reduce the size of the gains of the H∞ con- the signals to command the inverter switches, by means of a
trollers while keeping suitable performance. space vector modulation technique [38]. A deadtime is also
Since conditions FEAS1 and FEAS2 are based on the taken into account to avoid short circuit in the inverter legs. The
quadratic stability [23] and FEAS3 and FEAS4 are based on synchronization with the grid voltage at the PCC is guaranteed
a less conservative concept of stability called poly-quadratic by a Kalman filter algorithm, as in [39], also programmed in
stability [37], we produce controllers that guarantee stability the DLL. The references for the grid currents, iα and iβ , are
even for arbitrary parametric variations.3 The theoretical con- also generated in the DLL allowing, by changes of phase and
tribution here is condition (25), which provides a bound for the amplitude, to simulate the injection of active or reactive powers
control gain (constructed using the slack variable G) in terms of into the grid.
two scalars μ0 and δ, both appearing linearly in the condition. First, we consider the strategy of minimizing the H∞ norm,
The conditions OPT1, OPT2, and OPT3 are convex opti- given by the problem OPT1 s.t. (20) with r = 1. This problem
mization problems, leading to the best values of the objective leads to an H∞ norm given by γ ∗ = 0.5163. Although the value
functions in a very efficient computational way [23]. It is worth of γ from OPT1 is low, the control gains with OPT1 are very
to mention that (22) limits the Lyapunov matrix, while (25) high, not producing viable results in PSIM simulation. Similar
does not impose bounds directly on the Lyapunov matrices, behaviors are obtained with the optimization condition OPT1
contributing to reduce the conservatism of the design. s.t. (23) with r = 1, indicating that only the minimization of the
Notice that from the simple point of view of the feasibility of H∞ norm does not produce viable controllers for this practical
LMIs, the condition FEAS3 is the least conservative. However, application.
the point is not only the existence of numerical solution for Since H∞ optimal controllers do not provide viable results,
conditions FEAS1 to FEAS4, but also their gains to be effective the first indication is to choose γ > γ ∗ , that is, to investigate H∞
in practical applications. suboptimal controllers. For a performance comparison, consider
To test the gains produced by the above LMI conditions, the conditions FEAS1 and FEAS3, with γ = 5 (suboptimal con-
simulations are shown out in the next section. These simulations trollers), tested for r = 0.999, r = 0.997, and r = 0.993. Con-
consider the converter connected to the grid, as in Fig. 1, with sider also the conditions OPT2 and OPT3, tested for γ = 5 and
parameters given in Table I, where the grid inductance is defined the same values r, to verify the effect of constraining the con-
by an interval, to express the uncertainty in this parameter. trol gains. Simulations of the closed-loop system in PSIM, with
the conditions FEAS1, FEAS3, OPT2, and OPT3 produce the
IV. SIMULATION RESULTS results summarized in Table II.
The simulations of the system in Fig. 1 are carried out in From Table II, first one can notice that although all the condi-
software PSIM, using the full-bridge inverter with IGBT, a dc tions are feasible (i.e., all of them produce a control gain), only
some of the controllers lead to stable results in PSIM simula-
3 For this case, one has that property 2) in Problem 1 is restated as ρ(k) ≤ tions, under limited control action. Clearly, one can see better
r k ρ(0) , 0 < r ≤ 1, where r is an upper bound for the decay rate of the results with the condition OPT3 (all controllers produced by this
trajectories of the closed-loop system. condition presented stable responses). This is due to the slack

Authorized licensed use limited to: Seoul National Univ of Science & Tech (SNUT). Downloaded on July 04,2023 at 07:54:21 UTC from IEEE Xplore. Restrictions apply.
6026 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 66, NO. 8, AUGUST 2019

TABLE II Finally, for a comparison with similar conditions for H∞ state


STABILITY LIMITS (γ = 5)
feedback controllers that were implemented in practice, two
other robust controllers are designed now. The H∞ controller
given by [31], with r = 0.994, γ = 100 and the controller given
by [32] with r = 0.49 and γ = 4, which uses only one resonant
at the fundamental frequency. Additionally, the robust discrete
linear quadratic regulator (DLQR) from [29] is also used to
have a comparison with another optimal controller. These con-
trol gains are given in Table IV, respectively, in the columns
TABLE III
GAINS TO γ = 5, r = 0.997 AND OPTIMIZING k z , k w , δ AND μ 0 WCPL, MMSRPMO, and MSPOM. To corroborate the prac-
tical viability of the proposed H∞ state feedback controllers,
experimental results will be shown in the next section.

V. EXPERIMENTAL RESULTS
The prototype in Fig. 4 is used to obtain the experimen-
tal results. This prototype has a nominal power of 5.4 kW,
uses a full-bridge inverter based on IGBTs (Semikron), and
a three-phase LCL filter with parameters in Table I. Hall ef-
fect sensors (LEM) are used for the measurements of currents
and voltages of the LCL filter, and the control law is imple-
mented in the DSP TMS320F28335 (Texas Instruments). As
mentioned in Section IV, the synchronization with the grid
voltage in the PCC is obtained by means of a Kalman filter
variables used together with the control gain limitation. The algorithm [39].
control gains produced by FEAS1, FEAS3, OPT2, and OPT3 First, a test for tracking of a sinusoidal reference for the
for the case γ = 5 and r = 0.997, shown in Table II, can be seen current iα was implemented, including variations of phase and
in Table III, indicating much lower gains with OPT3. amplitude. The results are shown in Fig. 5, where the wave-
For a more detailed comparison, time simulation results with forms are obtained directly from the DSP data. As indicated
the gains from OPT2 and OPT3 in Table III are given in Fig. 2, by the ITAE measures from the simulation results in Fig. 3(b),
confirming that only OPT3 guarantees stable behavior, with a the experimental results in Fig. 5 confirm that the better per-
control action without saturation. It is worth to mention that the formance is provided by condition FEAS4. Then, the controller
results for FEAS1 and FEAS3 are unstable, and similar to those from FEAS4, with gains in Table IV, will be used for all the
in Fig. 2(a). other tests shown in this section.
Moreover, concerning previous similar works in the literature, The robust stability of the closed-loop system with the con-
it is important to point out that the design condition in [33] is troller from FEAS4 is confirmed by Fig. 6(a), which shows that
recovered here by FEAS3, and the design condition in [40] is the closed-loop eigenvalues for the entire range of Lg 2 given
particularized only for symmetric matrices G in (23). in Table I respect the constraint of r = 0.994. Moreover, since
Thus, OPT3 is a more efficient design condition, as confirmed the condition FEAS4 is based on a poly-quadratic Lyapunov
by the better performances in the simulations (see Fig. 2 and function, the controller ensures stability even under arbitrarily
Table II). A more detailed study is now carried out on the role of fast variations of parameter Lg 2 . Additionally, concerning ro-
r, γ, δ, and μ0 in the control gains and performance produced bust performance, Fig. 6(b) shows that the closed-loop system
by this condition. can track sinusoidal references with frequency of 60 Hz (0 dB
Fig. 3(a) shows the effect of γ and r in the norm of the control gain and phase equivalent to zero degrees). Fig. 6(c) confirms
gains. It is clear that lower values of γ and r lead to higher that the closed-loop system can reject grid voltage harmonics at
control gains, which can represent difficulties for practical the frequencies of the resonant controllers. One can also notice
implementation. that the peak of the Bode diagram of Fig. 6(c) respects the limit
In order to improve the quality of the grid-injected current of γ = 5 (or 6.9897 dB), which can be seen as a limit of the
waveform, in Fig. 3(b), the integral time absolute error (ITAE) output admittance of the converter, as detailed in the following
criterion was computed in the time interval with colored back- remark.
ground in Fig. 2(b), for comparison of the H∞ optimal con- Remark 1: The output admittance of the converter, seen by
troller OPT3 with a H∞ suboptimal controller FEAS4. The the PCC, can be obtained by replacing the third equation of (8)
gains for these controllers can be seen in Table IV, OPT3 and (9) by
(r = 0.994, γ = 5, minimizing δ − μ0 ) and FEAS4 (r = 0.994,
γ = 5, choosing δ = 5.5 and μ0 = 0.00013), confirming the
dig 1 1
superior performance in terms of ITAE with the relaxations of = vc − vPCC (26)
δ and μ0 , allowed by the condition FEAS4. dt Lg 1 Lg 1

Authorized licensed use limited to: Seoul National Univ of Science & Tech (SNUT). Downloaded on July 04,2023 at 07:54:21 UTC from IEEE Xplore. Restrictions apply.
KOCH et al.: ROBUST H∞ STATE FEEDBACK CONTROLLERS BASED ON LMIs APPLIED TO GRID-CONNECTED INVERTERS 6027

Fig. 2. Results for the closed-loop simulation in PSIM for variations in phase and amplitude of the sinusoidal reference using conditions OPT2 and
OPT3: (a) Grid current and its reference, and respective control signal u α (divided by V c c ) for OPT2 (with γ = 5 and r = 0.997). (b) Grid current
and its reference, and respective control signal u α (divided by V c c ) for OPT3 (with γ = 5 and r = 0.997).

Fig. 3. (a) Vertical axis: norm-2 of the control gains; horizontal axis:
value of γ, for three different cases of r, and optimizing δ and μ 0 , for
OPT3. (b) Vertical axis: ITAE; horizontal axis: OPT3, defined by r =
0.994, γ = 5 and optimizing δ and μ 0 ; FEAS4 defined by r = 0.994,
γ = 5, δ = 5.5 and μ 0 = 0.00013.

TABLE IV
COMPARISON OF CONTROL GAINS

Fig. 5. Experimental results for OPT3 and FEAS4 with controllers in


Table IV: (a) iα and iref , (b) tracking error for the interval marked by (1) in
Fig. 5(a), and (c) tracking error for the interval marked by (2) in Fig. 5(a).

which points out that the frequency response from vPCC to ig is


actually independent of Lg 2 , and is equal to the curve marked as
Lg 2min (i.e., 0 mH) in Fig. 6(b). Then, the H∞ norm, γ, chosen
by the control designer when using FEAS4, is an upper bound
for the admittance of the converter, seen from the PCC, for the
entire frequency range.
Fig. 7 shows the tracking of sinusoidal references in axes
α and β, with changes of phase and amplitude. One can notice
good steady-state responses and also suitable transients, in terms
of overshoots and settling times. From the design parameter r,
one can expect the transient response modes will decay at most
(for the criterion of 2% of error) in two cycles of the reference.
In Fig. 8, one can confirm the good correspondence between
simulation and experimental results, indicating the suitability of
the models and conditions used here for control design.
The three-phase currents injected into the grid, corresponding
to the currents in α and β shown in Fig. 7, can be seen in Fig. 9,
confirming good transient and steady-state performances.
Fig. 10 shows a detailed view of the three-phase grid cur-
rents, together with the harmonic components for one of the
Fig. 4. Prototype used for the experimental results. waveforms, respecting the limits of the IEEE Standard 1547.

Authorized licensed use limited to: Seoul National Univ of Science & Tech (SNUT). Downloaded on July 04,2023 at 07:54:21 UTC from IEEE Xplore. Restrictions apply.
6028 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 66, NO. 8, AUGUST 2019

Fig. 8. Detailed comparison of the experimental and simulation results


for controller FEAS4, in Table IV (for L g max ).

Fig. 6. Analyses for controller FEAS4, in Table IV: (a) Closed-loop


eigenvalues for L g ∈ [L g m in , L g m a x ]. (b) and (c) Bode diagram of the
closed-loop system from the input iref to the output ig and from the input
v g to the output ig , respectively.

Fig. 9. Experimental result: three-phase grid currents for the closed-


loop system with controller FEAS4, in Table IV.

Fig. 7. Experimental closed-loop results for references with changes Fig. 10. Experimental results for controller FEAS4, in Table IV: steady-
in phase and amplitude for controller FEAS4, in Table IV: (a) iα . (b) iβ . state grid currents and harmonic content of the waveform marked
with the arrow, respecting the limits from IEEE Standard 1547, with
THD = 2.61%.
The THD of this waveform is 2.61% (<5%). Thus, this current
complies with the standard requirements. The same conclusion
is valid for the currents in the other phases. results with H∞ controller from FEAS4, and the unsuitable
Experimental short circuit tests for a sudden variation of the performance with the H∞ controller from WCPL [31]. Similar
grid inductance are provided in Fig. 11. System starts opera- unsuitable performance is also verified for the controller from
tion with Lg 2 = Lg 2min and at the instant pointed by the arrow, [32] in this test. The H∞ controllers from [31] and [32] also
Lg 2 is suddenly changed to Lg 2max . One can notice the good do not provide acceptable results when connected to the grid,

Authorized licensed use limited to: Seoul National Univ of Science & Tech (SNUT). Downloaded on July 04,2023 at 07:54:21 UTC from IEEE Xplore. Restrictions apply.
KOCH et al.: ROBUST H∞ STATE FEEDBACK CONTROLLERS BASED ON LMIs APPLIED TO GRID-CONNECTED INVERTERS 6029

condition FEAS4, provide experimental results complying with


the limits prescribed by the IEEE Standard 1547, confirming
the practical viability of this technique as an alternative for this
application.

APPENDIX
Since the conditions FEAS1 and FEAS3 can deal with time-
varying parameters, the proofs are presented considering this
more general scenario. For the proof of FEAS1, one has that,
replacing Z = KW in (20) multiplying each inequality by θi (k)
and summing up for i = 1, 2 one has
⎡  ⎤
G(θ(k)) Hu (θ(k))
W + K W 0 Hd (θ(k))
⎢ r r ⎥
⎢ ⎥
⎢ W WCaug
0 ⎥
⎢ ⎥
⎢ ⎥
⎢ I 0 ⎥
Fig. 11. Experimental results for a sudden variation of the output induc-
⎣ ⎦
2
tance from L g 2 min to L g 2 max : (a) output current and error for controller γ I
FEAS4, in Table IV, (b) output current and error for controller WCPL, in
Table IV. > 0. (27)

Multiplying (27) on the left-hand side and on the right-hand


side by T , with
⎡ −1 ⎤
W 0 0 0
⎢ ⎥
⎢ W −1 0 0 ⎥
T =⎢ ⎢ ⎥
0 ⎥
⎣ I ⎦
I

and replacing P = W −1 , one has


⎡ ⎤
P PAcl (θ(k)) 0 PHd (θ(k))
⎢  ⎥
⎢ P Caug 0 ⎥
Fig. 12. Experimental closed-loop results for references with changes ⎢ ⎥>0 (28)
⎢ 0 ⎥
in phase and amplitude for controllers FEAS4 and MSPOM, in Table IV. ⎣ I ⎦
γ2 I
due to limitation of the control signal in the startup, which is
with Acl (θ(k)) = ( G (θr(k )) + H u (θr (k )) K). Multiplying (28) on
overcome by the H∞ controller from FEAS4. 

Finally, a comparison of the proposed H∞ controller with an- the left-hand side by R and on the right-hand side by R , with
other optimal robust controller in the literature, the robust DLQR ⎡ ⎤
0 I 0 0
from [29] (gain MSPOM in Table IV), is given in Fig. 12. One ⎢ ⎥
⎢0 0 0 I ⎥
can see by the experimental results of this test that both con- R=⎢ ⎢0 0 I 0⎥

trollers have similar performances in steady state, with advan- ⎣ ⎦
tage for the H∞ controller in the transient responses. I 0 0 0

VI. CONCLUSION provides


⎡ 

This paper showed the viability of an H∞ state feedback con- P 0 Caug Acl (θ(k)) P
⎢ ⎥
troller, robust to parameter uncertainties and variations for three- ⎢ γ2 I 0 Hd (θ(k)) P ⎥
phase grid-connected converters. The proposed design condition ⎢ ⎥ > 0. (29)
⎢ 0 ⎥
was used in FEAS4, based on LMIs relying on extra variables to ⎣ I ⎦
limit the control gains without bounding directly the Lyapunov P
matrices, which allows less conservative results. This condi-
Applying a Schur complement, yields
tion allowed the control designer to choose, for instance, an
 
upper bound for the converter output admittance. Comparisons P 0 I 0 
with other H∞ state feedback controllers and other robust op- 2
>Q −1
Q
timal controller in the literature illustrate the superiority of the γ I 0 P

Authorized licensed use limited to: Seoul National Univ of Science & Tech (SNUT). Downloaded on July 04,2023 at 07:54:21 UTC from IEEE Xplore. Restrictions apply.
6030 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 66, NO. 8, AUGUST 2019

with Applying the following transformation:


   
 
Caug Acl (θ(k)) P G −1

0 G  (μ0 In )−1 G F  G −1 0
Q=  >0
0 Hd (θ(k)) P Im δIm Im
which leads to one has
 
Acl (θ(k)) PAcl (θ(k)) − P + Caug

Caug M12 (μ0 In )−1

G −1 F 
< 0 (30) > 0.
M22 δIm
where Applying Schur complement, it is possible to write
  
M12 = Acl (θ(k)) PHd (θ(k)) δIm > FG −1 μ0 In G −1 F  ⇒ δμ−1
0 Im > FG
−1 −1
G F
M22 = Hd (θ(k)) PHd (θ(k)) − γ 2 I. ⇒ δμ−1 
0 Im > KK .

Multiplying (30) on the left-hand side by [ρ(k) w(k) ] and on Hence, KK  < δμ−1 2
0 Im . For Im = 1, K2 < δ/μ0 .
the right-hand side by its transpose, and taking into account
that ρ(k + 1) = Acl (θ(k))ρ(k) + Hd (θ(k))w(k) and y(k) = REFERENCES
Caug ρ(k), one has
[1] R. Teodorescu, M. Liserre, and P. Rodrı́guez, Grid Converters for Photo-
ρ(k + 1) Pρ(k + 1) − ρ(k) Pρ(k) + y(k) y(k) voltaic and Wind Power Systems. New York, NY, USA: Wiley, 2011.
[2] F. Blaabjerg and K. Ma, “Future on power electronics for wind turbine
− γ 2 w(k) w(k) < 0. systems,” IEEE J. Emer. Sel. Topics Power Electron., vol. 1, no. 3, pp. 139–
152, Sep. 2013.
[3] F. Blaabjerg, Z. Chen, and S. Kjaer, “Power electronics as efficient inter-
Summing up the left-hand side from k = 0 to k = ∞, one has face in dispersed power generation systems,” IEEE Trans. Power Electron.,
∞ vol. 19, no. 5, pp. 1184–1194, Sep. 2004.
ρ(∞) Pρ(∞) − ρ(0) Pρ(0) + y(k) y(k) [4] S. Buso and P. Mattavelli, Digital Control in Power Electronics. San
Rafael, CA, USA: Morgan & Claypool, 2006.
k =0 [5] IEEE Standard for Interconnecting Distributed Resources With Electric
∞ Power Systems, Institute of Electrical and Electronics Engineers Std.,
− γ2 w(k) w(k). IEEE 1547-2003, 2011.
[6] F. Blaabjerg, R. Teodorescu, M. Liserre, and A. Timbus, “Overview of
k =0 control and grid synchronization for distributed power generation sys-
tems,” IEEE Trans. Ind. Electron., vol. 53, no. 5, pp. 1398–1409, Oct.
Applying the definition of norm-2 of a signal, it follows that 2006.
ρ(∞) Pρ(∞) − ρ(0) Pρ(0) + y(k)22 − γ 2 w(k)22 < 0. [7] M. Liserre, R. Teodorescu, and F. Blaabjerg, “Stability of photovoltaic and
wind turbine grid-connected inverters for a large set of grid impedance
values,” IEEE Trans. Power Electron., vol. 21, no. 1, pp. 263–272, Jan.
From the block (1,1) in (30), one has that matrix Acl (θ(k)) 2006.
is robustly stable and that r is a upper bound for the decay [8] W. Wu, Y. Liu, Y. He, H. S. H. Chung, M. Liserre, and F. Blaabjerg,
rate of the trajectories of the closed-loop system (eigenvalues “Damping methods for resonances caused by LCL-filter-based current-
controlled grid-tied power inverters: An overview,” IEEE Trans. Ind.
inside a circle with radius r in the time-invariant case). Finally, Electron., vol. 64, no. 9, pp. 7402–7413, Sep. 2017.
assuming zero initial conditions (i.e., ρ(0) = 0), one has that [9] S. Jayalath and M. Hanif, “Generalized LCL-filter design algorithm
limk →∞ ρ(k) = 0 and, as a consequence for grid-connected voltage-source inverter,” IEEE Trans. Ind. Electron.,
vol. 64, no. 3, pp. 1905–1915, Mar. 2017.
y(k)2 < γw(k)2 (31) [10] Z. Xin, X. Wang, P. C. Loh, and F. Blaabjerg, “Grid-current-feedback
control for LCL-filtered grid converters with enhanced stability,” IEEE
which ensures that γ provides an upper bound for the H∞ norm Trans. Power Electron., vol. 32, no. 4, pp. 3216–3228, Apr. 2017.
[11] R. Peña-Alzola, M. Liserre, F. Blaabjerg, R. Sebastián, J. Dannehl, and F.
of the system. W. Fuchs, “Analysis of the passive damping losses in LCL-filter-based grid
For the proof of FEAS2, applying Schur complement to the converters,” IEEE Trans. Power Electron., vol. 28, no. 6, pp. 2642–2646,
inequalities given in (22), one has, respectively [23] Jun. 2013.
[12] R. N. Beres, X. Wang, F. Blaabjerg, M. Liserre, and C. L. Bak, “Opti-
W −1 < kw I, Z  Z < kz I. mal design of high-order passive-damped filters for grid-connected ap-
plications,” IEEE Trans. Power Electron., vol. 31, no. 3, pp. 2083–2098,
Mar. 2016.
Replacing this expression in K  K, it is possible to write [13] A. Timbus, M. Liserre, R. Teodorescu, P. Rodriguez, and F. Blaab-
jerg, “Evaluation of current controllers for distributed power generation
(ZW −1 ) (ZW −1 ) = W −1 Z  ZW −1 < kz IW −1 W −1 < kz kw2 I systems,” IEEE Trans. Power Electron., vol. 24, no. 3, pp. 654–664,
Mar. 2009.
and, hence, K  K < kz kw2 I. [14] J. Dannehl, F. Fuchs, and P. Thøgersen, “PI state space current control
Proof of FEAS3 can be derived from [37, Th. 4]. of grid-connected PWM converters with LCL filters,” IEEE Trans. Power
For the proof of FEAS4, notice that for μ0 > 0, one has (G − Electron., vol. 25, no. 9, pp. 2320–2330, Sep. 2010.
[15] X. Wang, F. Blaabjerg, and P. C. Loh, “Grid-current-feedback active damp-
μ0 In ) (μ0 In )−1 (G − μ0 In ) ≥ 0 and, hence, G  (μ0 In )−1 G ≥ ing for LCL resonance in grid-connected voltage-source converters,” IEEE
G + G  − μ0 In . Applying this bound to (25), one has Trans. Power Electron., vol. 31, no. 1, pp. 213–223, Jan. 2016.
 [16] M. Liserre, R. Teodorescu, and F. Blaabjerg, “Multiple harmonics control
G  (μ0 In )−1 G F  for three-phase grid converter systems with the use of PI-RES current
> 0. controller in a rotating frame,” IEEE Trans. Power Electron., vol. 21,
δIm no. 3, pp. 836–841, May 2006.

Authorized licensed use limited to: Seoul National Univ of Science & Tech (SNUT). Downloaded on July 04,2023 at 07:54:21 UTC from IEEE Xplore. Restrictions apply.
KOCH et al.: ROBUST H∞ STATE FEEDBACK CONTROLLERS BASED ON LMIs APPLIED TO GRID-CONNECTED INVERTERS 6031

[17] R. Teodorescu, F. Blaabjerg, M. Liserre, and P. Loh, “Proportional- [39] R. Cardoso, R. F. de Camargo, H. Pinheiro, and H. A. Gründling, “Kalman
resonant controllers and filters for grid-connected voltage-source con- filter based synchronisation methods,” IET Gener., Transmiss. Distrib.,
verters,” IEE Proc. Electr. Power Appl., vol. 153, no. 5, pp. 750–762, vol. 2, no. 4, pp. 542–555, Jul. 2008.
Sep. 2006. [40] G. G. Koch, H. Pinheiro, L. A. Maccari, R. C. L. F. Oliveira, A. A.
[18] J. Vasquez, J. Guerrero, M. Savaghebi, J. Eloy-Garcia, and R. Teodorescu, Ferreira, and V. F. Montagner, “Robust H∞ control for rejection of voltage
“Modeling, analysis, and design of stationary-reference-frame droop- disturbances in grid-connected converters,” in Proc. 12th IEEE Int. Conf.
controlled parallel three-phase voltage source inverters,” IEEE Trans. Ind. Ind. Appl., Curitiba, PR, Brazil., Nov. 2016, pp. 1–6.
Electron., vol. 60, no. 4, pp. 1271–1280, Apr. 2013.
[19] D. Pan, X. Ruan, C. Bao, W. Li, and X. Wang, “Optimized controller
design for LCL-type grid-connected inverter to achieve high robustness
against grid-impedance variation,” IEEE Trans. Ind. Electron., vol. 62,
no. 3, pp. 1537–1547, Mar. 2015. Gustavo Guilherme Koch (S’16) was born in
[20] M. Hanif, V. Khadkikar, W. Xiao, and J. Kirtley, “Two degrees of free- Ijuı́, Rio grande do Sul, Brazil, in 1988. He re-
dom active damping technique for LCL filter-based grid connected PV ceived the B.Sc. and M.Sc. degrees in electrical
systems,” IEEE Trans. Ind. Electron., vol. 61, no. 6, pp. 2795–2803, engineering in 2013 and 2015, respectively, from
Jun. 2014. the Federal University of Santa Maria, Santa
[21] J. Xu, S. Xie, and T. Tang, “Active damping-based control for grid- Maria, Brazil, where he is currently working to-
connected LCL-filtered inverter with injected grid current feedback ward the Ph.D. degree with the Power Electron-
only,” IEEE Trans. Ind. Electron., vol. 61, no. 9, pp. 4746–4758, Sep. ics and Control Research Group.
2014. His research interests include robust con-
[22] J. Dannehl, M. Liserre, and F. Fuchs, “Filter-based active damping of trol and control theory applications on power
voltage source converters with LCL filter,” IEEE Trans. Ind. Electron., electronics.
vol. 58, no. 8, pp. 3623–3633, Aug. 2011.
[23] S. Boyd, L. El Ghaoui, E. Feron, and V. Balakrishnan, Linear Matrix
Inequalities in System and Control Theory. Philadelphia, PA, USA: SIAM,
1994.
[24] E. Feron, P. Apkarian, and P. Gahinet, “Analysis and synthesis of ro-
bust control systems via parameter-dependent Lyapunov functions,” IEEE Luiz Antonio Maccari Jr. was born in Sarandi,
Trans. Autom. Control, vol. 41, no. 7, pp. 1041–1046, Jul. 1996. RS, Brazil, in 1985. He received the B.Sc.,
[25] J. Daafouz and J. Bernussou, “Parameter dependent Lyapunov functions M.Sc., and Ph.D. degrees in electrical engineer-
for discrete time systems with time varying parameter uncertainties,” Syst. ing from the Federal University of Santa Maria,
Control Lett., vol. 43, no. 5, pp. 355–359, Aug. 2001. Santa Maria, Brazil, in 2009, 2011, and 2015,
[26] M. C. de Oliveira, J. C. Geromel, and J. Bernussou, “Extended H2 and respectively.
H∞ characterization and controller parametrizations for discrete-time sys- From 2014 to 2015, he was a Professor with
tems,” Int. J. Control, vol. 75, no. 9, pp. 666–679, 2002. Federal University of Pampa, Bage, Brazil. Since
[27] L. A. Maccari et al., “LMI-based control for grid-connected converters 2015, he has been a Professor with Federal Uni-
with LCL filters under uncertain parameters,” IEEE Trans. Power Elec- versity of Santa Catarina, Florianopolis, Brazil,
tron., vol. 29, no. 7, pp. 3776–3785, Jul. 2014. where he teaches on dynamical systems and
[28] I. Gabe, V. F. Montagner, and H. Pinheiro, “Design and implementation control theory. His research interests include robust control techniques,
of a robust current controller for VSI connected to the grid through an linear matrix inequalities and control theory applications on power elec-
LCL filter,” IEEE Trans. Power Electron., vol. 24, no. 6, pp. 1444–1452, tronics, renewable energy systems, and electrical machines.
Jun. 2009.
[29] L. A. Maccari, Jr, A. C. L. Santini, H. Pinheiro, R. C. L. F. de Oliveira, and
V. F. Montagner, “Robust optimal current control for grid-connected three-
phase pulse-width modulated converters,” IET Power Electron., vol. 8,
no. 8, pp. 1490–1499, Mar. 2015.
[30] K. Zhou, J. C. Doyle, and K. Glover, Robust and Optimal Control. Upper Ricardo C. L. F. Oliveira was born in Assis
Saddle River, NJ, USA: Prentice-Hall, 1996. Chateaubriand, Paraná, Brazil, in 1978. He re-
[31] G. Willmann, D. F. Coutinho, L. F. A. Pereira, and F. B. Libano, “Multiple- ceived the B.Sc. degree in computer engineering
loop H∞ control design for uninterruptible power supplies,” IEEE Trans. from the Pontifical Catholic University of Paraná,
Ind. Electron., vol. 54, no. 3, pp. 1591–1602, Jun. 2007. Curitiba, Brazil, in 2001, and the M.Sc. and
[32] L. A. Maccari et al., “Robust H∞ control for grid connected PWM in- Ph.D. degrees in electrical engineering from the
verters with LCL filters,” in Proc. 10th IEEE/IAS Int. Conf. Ind. Appl., University of Campinas (UNICAMP), Campinas,
Fortaleza, CE, Brazil, Nov. 2012, pp. 1–6. Brazil, in 2003 and 2006, respectively.
[33] G. G. Koch, H. Pinheiro, R. C. L. F. Oliveira, and V. F. Montagner, He is currently a Professor with the School
“Experimental evaluation of robust and nonrobust H∞ controllers for of Electrical and Computer Engineering, UNI-
three-phase grid-connected converters,” in Proc. Brazilian Power Electron. CAMP, Brazil. His research interests include the
Conf., Juiz de Fora, MG, Brazil., Nov. 2017, pp. 1–6. development of numerical tools for stability analysis, filtering and control
[34] W. Duesterhoeft, M. W. Schulz, and E. Clarke, “Determination of instan- design of uncertain linear, LPV, and fuzzy systems.
taneous currents and voltages by means of alpha, beta, and zero com-
ponents,” Trans. Am. Inst. Electr. Eng., vol. 70, no. 2, pp. 1248–1255,
Jul. 1951.
[35] B. A. Francis and W. M. Wonham, “The internal model principle of control
theory,” Automatica, vol. 12, no. 5, pp. 457–465, Sep. 1976.
Vinı́cius Foletto Montagner received the Ph.D.
[36] L. Pereira, J. Flores, G. Bonan, D. Coutinho, and J. da Silva, “Multiple
degree in electrical engineering from the Univer-
resonant controllers for uninterruptible power supplies—A systematic ro-
bust control design approach,” IEEE Trans. Ind. Electron., vol. 61, no. 3, sity of Campinas, Campinas, Brazil, in 2005.
He is currently a Professor with the Federal
pp. 1528–1538, Mar. 2014.
University of Santa Maria, Santa Maria, Brazil,
[37] J. Daafouz and J. Bernussou, “Poly-quadratic stability and H∞ perfor-
and works with the Power Electronics and Con-
mance for discrete systems with time varying uncertainties,” in Proc. 40th
IEEE Conf. Decis. Control, Orlando, FL, USA, 2001, vol. 1, pp. 267–272. trol Research Group. His research interests in-
clude control theory and applications.
[38] D. G. Holmes and T. A. Lipo, Pulse Width Modulation for Power Con-
verters: Principles and Practice. Piscataway, NJ, USA: Wiley-IEEE Press,
2003.

Authorized licensed use limited to: Seoul National Univ of Science & Tech (SNUT). Downloaded on July 04,2023 at 07:54:21 UTC from IEEE Xplore. Restrictions apply.

You might also like