IEEE JOURNAL OF EMERGING AND SELECTED TOPICS IN POWER ELECTRONICS, VOL. 5, NO.

4, DECEMBER 2017 1455

AC Stability Analysis and dq Frame Impedance

Specifications in Power-Electronics-Based
Distributed Power Systems
Bo Wen, Member, IEEE, Rolando Burgos, Member, IEEE, Dushan Boroyevich, Fellow, IEEE,
Paolo Mattavelli, Fellow, IEEE, and Zhiyu Shen, Member, IEEE

Abstract— Small-signal stability in balanced three-phase sys-

tems is typically investigated by means of the generalized Nyquist
stability criterion (GNC) that involves operations on the source
and load subsystems’ impedance matrices. Eigenvalues of the
ratio of these impedance matrices should be calculated for stabil-
ity judgment. This paper shows that for a power-electronics-based
distributed power system, impedance matrices of subsystems can
be designed as diagonal dominant. Therefore, stability, in this
case, is fully determined by the scalar ratios formed by the
impedance elements seen across the d-axis and q-axis interfaces. Fig. 1. Power-electronics-based DPS.
A three-phase ac system can then be treated as two decoupled
dc systems. As a result, the ac stability analysis can also be
conducted based on insightful impedance quantities; impedance among all these DPSs is the use of power electronics as
specification criteria developed for dc systems can be readily the means to interface sources and loads as they are shown
applied to ensure stability in ac systems. The major contribution in Fig. 1. In this way, the DPSs feature with high efficiency
of this paper is related to the simplification of the GNC analysis
with good experimental validations. Limitation of the proposed and high power factor (PF). Usually, a DPS is not designed
method is that the systems have to work under high power factor as a whole due to its complexity: it is typically defined in
condition. However, such limitation is not a problem for many terms of several subsystems, and each subsystem is designed
power-electronics-based distributed power systems. and manufactured individually by different suppliers. The
Index Terms— Impedance, inverters, power system stability, subsystems are then integrated to form a complete DPS by
rectifiers, stability. the system integrator. One of the most important problems
involved with system integration is the possible instability
I. I NTRODUCTION
due to subsystem interactions. Even though each subsystem

T HE electrification process that aviation, maritime, and

ground transportation applications [1], [2] have gone
through over the past years is a direct result of energy
may be designed properly for idealized source and load
operation, the DPS could become unstable after the sub-
systems are integrated due to interconnections of multiple
conservation and sustainability. Other applications pursuing tightly regulated power converters [3]–[9]. More recently, the
high-efficiency targets are data centers and microgrids. As a use of grid-tied inverters (GTIs) in microgrid and renewable
result, distributed power systems (DPSs) are widely used in energy applications has revealed interactions in ac distribution
the aforementioned applications. The common denominator systems [15]–[19].
Many techniques, such as eigenvalues analysis based on
Manuscript received January 31, 2017; revised May 9, 2017; accepted
July 3, 2017. Date of publication July 18, 2017; date of current version Octo- system state-space [20], loop-gain [21]–[23], and passiv-
ber 30, 2017. This work was supported by the Center for Power Electronics ity [24], [25], impedance-based method [4], [10], are pro-
Systems Renewable Energy and Nanogrids mini-consortium. Recommended posed to study the system stability problem. A preferred
for publication by Associate Editor Marta Molinas. (Corresponding author:
Bo Wen.) approach due to its ease of practical implementation has
B. Wen is with the School of Electrical and Electronic Engineer- been impedance-based analysis, which has been favored too,
ing, The University of Manchester, Manchester M15 4TB, U.K. (e-mail: given its inherent encrypted nature where vendors can openly
[Link]@[Link]).
R. Burgos and D. Boroyevich are with the Center for Power Electronics Sys- share the terminal characteristics of their equipment aiding
tems, Bradley Department of Electrical and Computer Engineering, Virginia in the design and integration of DPSs. It has been used
Tech, Blacksburg, VA 24061 USA (e-mail: rolando@[Link]; dushan@[Link]). to investigate input–filter–converter interactions [4], cascaded
P. Mattavelli is with the University of Padua, 36100 Vicenza, Italy (e-mail:
[Link]@[Link]). dc–dc converters [5]–[7], cascaded ac–ac converters and
Z. Shen is with GE Global Research, Niskayuna, NY 12309 USA (e-mail: ac interfaces in general [9]–[12], harmonic stability [13], [14],
zhiyu@[Link]). and GTI subsynchronous interactions [15]–[19]. For dc sys-
Color versions of one or more of the figures in this paper are available
online at [Link] tems, as stipulated originally by Middlebrook [4], this
Digital Object Identifier 10.1109/JESTPE.2017.2728640 approach simply uses the Nyquist plots of the impedance ratio
2168-6777 © 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See [Link] for more information.
1456 IEEE JOURNAL OF EMERGING AND SELECTED TOPICS IN POWER ELECTRONICS, VOL. 5, NO. 4, DECEMBER 2017

seen at any given interface to assess stability. Accordingly, II. AC S YSTEM S MALL -S IGNAL S TABILITY A NALYSIS
stability margins can easily be formulated by defining for- A. Small-Signal Stability Assessment
bidden regions for the locus of the impedance ratio [5]–[7]
1) dq Domain: By using dq transformation [12], [18],
or alternatively transformed into load impedance specifica-
time-varying variables of balanced three-phase systems are
tions using Bode plots [5]. Impedance specification is a very
transferred to the dq domain to generate a dc operating
effective and convenient technique to integrate new load to
point for linearization purposes. Then, similar to a dc system,
a DPS. Integrator of the DPS can specify the shape of the input
ac systems in the dq domain can be modeled by a source
impedance of the new load according to the output impedance
and load equivalent representation. Zsdq is the source output
of the source in the DPS to secure the system stability [5].
impedance, and Yldq is the load input admittance, defined as
Supplier of the load can shape the input impedance of the
the inverse of the load input impedance matrix Zldq as follows:
load by using various control techniques [26] without exposing
−1
their internal design information. This method can greatly Yldq (s) = Zldq (s) (1)
facilitate the system integration and subsystem design of a
dc DPS. where upper case bold font variables are used to repre-
Although the stability at ac interfaces can be readily pre- sent matrices. These impedances are 2 × 2 matrices in the
dicted by the generalized Nyquist stability criterion (GNC) dq domain as shown in the following, whose elements Z ij (s)
using measured impedances in either dq or sequence describe the voltage response in the i (d or q)-axis to
domain [12], [27], [28], similar impedance specifications as a j (d or q)-axis current perturbation [12]
 
the ones formulated for dc systems have not been devel- Z dd (s) Z dq (s)
Zdq (s) = . (2)
oped so far. This is due primarily to the fact that stability Z qd (s) Z qq (s)
does not depend directly on the impedances seen at a
given interface—as in the dc case, but depends instead on Stability at the ac interface can then be assessed applying
the eigenvalues of the return-ratio matrix, or even more the GNC to the corresponding return-ratio matrix [12]
involved, on its determinant, which makes this relationship less L(s) = Zsdq (s)Yldq (s)
clear and harder to grasp. In consequence, existent methods    
Z sdd (s) Z sdq (s) Y (s) Yldq (s)
rely on the small-gain stability criterion to limit or shape = · ldd . (3)
Z sqd (s) Z sqq (s) Ylqd (s) Ylqq (s)
ac impedances [9]–[11]. This approach results in excessive
conservatism, similar to Middlebrook’s initial proposal for dc The reader is referred to [9] for its detailed description.
systems [4], and as such has not been very practical. So, Specifically, if {λ1 (s), λ2 (s)} is the set of frequency-dependent
a decoupled impedance model for an ac system is highly eigenvalues of the 2 × 2 return-ratio matrix L(s), defined by
appealing, because then the ac system can be simplified the following characteristic polynomial in λ (s) :
to two single-input-single-output (SISO) systems that not (λ(s)) ≡ det(λ(s)I − L(s)) = 0. (4)
only makes the analysis very simple but also greatly facil-
itate the system integration and subsystem design of the According to the GNC, the stability of the ac system relies
ac DPSs. on the positions of λ(s) on the complex plane which involves
Circumventing the inherent barrier posed by ac systems, the the calculation of the determinant of L(s), not directly the
contributions of this paper show that in the presence of high impedances of source and load.
PF loads and well regulated, high power quality sources— 2) Sequence Domain: In [10], sequence-domain imped-
both inherent to power-electronics-based DPSs, the dynamic ances, see (5), are defined using harmonic linearization,
interactions at ac interfaces can be effectively reduced to those i p and i n are the current perturbations in positive and negative
occurring through the d-axis and q-axis channels separately. sequences, v p and v n are the voltage responses accordingly.
Such dynamic decoupling, the property of closed-loop con- Z p is denoted the positive sequence impedance, and Z n is
trolled pulse width modulation (PWM) converters [30], [31], denoted the negative sequence impedance
enables the use of stability margins to do load impedance
Z p = v p /i p , Z n = v n /i n . (5)
specifications for ac systems. Specifically, in this paper,
the 6-dB gain and 60° phase margins are used to specify the In [27], couplings between positive sequence and negative
shapes of Z dd and Z qq of the load converter. Based on the sequence are found due to the phase-locked loop (PLL) effect
compatibility of the Z dd and Z qq to the specifications, the sys- for a GTI. Without considering the couplings, stability analysis
tem stability is predicted. System oscillation can be mitigated using Z p and Z n can lead to a wrong conclusion. Further,
by shaping Z dd and Z qq within the specifications. in [28], such couplings are identified in DPS components
This paper is organized as follows. Section II discusses where a PLL, a dc-link voltage controller, or unsymmetrical
the impedance-based stability analysis for ac system and the controllers are used. The modified sequence impedance matrix
idea of decoupling the system for impedance specification. is given in [28] as
Section III delves into the ac impedance of power converters to  
Z pp (s) Z pn (s)
show that in dq-domain, impedance matrix can be decoupled Z pn (s) = . (6)
Z np (s) Z nn (s)
for high PF systems. Section IV presents the experimental
results used to validate the proposed method, and the conclu- For stability analysis, GNC should be applied to the
sion is drawn in Section V. new return-ratio matrix defined by sequence impedance
WEN et al.: AC STABILITY ANALYSIS AND dq FRAME IMPEDANCE SPECIFICATIONS 1457

matrices [27], [28]. The procedure is the same as it is in

the dq domain [28]. Stability itself ultimately depends on
the eigenvalues of the return-ratio matrix and not directly
on the source and load impedance ratio as in the dc system
case.

B. Simplification to SISO Systems

As discussed, the main difficulty encountered when develop-
ing impedance specification criteria for ac power systems is the
inherent coupled dynamic nature that they exhibit as multi-in
multi-out (MIMO) systems. In the dq-domain, this is primarily
the result of coupling introduced by reactive components
(ac cables and filters), by relatively low bandwidth and
Fig. 2. Power stage and control system of AFE converter.
nondecoupled controllers, and by low PF operating condi-
tions. Under such power system characteristics, the return-
ratio matrix L(s) at ac interfaces will take the form in (4), Once the impedance matrices are diagonal, it is easy to link
since all source and load matrix elements Z ij (s) will have the parameters of the controllers to the shape of Z dd and Z qq .
comparable magnitude. If, however, an ac distribution system In the following section, dq-domain impedance matrices of
with high penetration of power electronics is considered, an active front end (AFE), a voltage-source inverter (VSI),
i.e., a DPS based on power electronics, then the coupled and a GTI are discussed to show their diagonal features when
characteristic mentioned above can be superseded by what are working on high PF condition.
essentially decoupled dq-domain dynamics. This is a result
of the high PF and low impedance characteristic featured by III. DQ-D OMAIN I MPEDANCES IN
power electronics loads and sources. Tightly regulated power D ISTRIBUTED P OWER S YSTEMS
converters now embody the sources and loads with decoupled
A. AFE Converter
dq-domain controllers, which actively control power flow and
minimize reactive power circulation in the system. An AFE converter such as that is shown in Fig. 2 act
Under the above conditions, the dq-domain impedances seen as an ac interface for a dc bus or dc-fed loads in the DPS.
at an ac interface will resemble that of a balanced set of Fig. 2 depicts specifically a two-level three-phase boost recti-
three-phase resistors [19], and will hence be characterized too fier with a dual-loop dq-coordinate controller synchronized to
by diagonally dominant matrices [19]. This will be detailed in the ac bus by means of a dq-coordinate PLL [34]. Although
Section III. The return-ratio matrix L(s) will in consequence a two-level topology is used as an example, the following
also assume a diagonally dominant structure as shown in (7), discussion applies to all types of three-phase AFE convert-
with its diagonal elements corresponding to the scalar return- ers. Further, an input LC filter is also depicted as these
ratio seen at the dd and qq interfaces, respectively are typically required for power quality and electromagnetic
interference (EMI) compliance assurance.
L(s) = Zsdq (s) · Yldq (s) One of the main features of this type of converter is
   
Z sdd (s) 0 Yldd (s) 0 PF correction (PFC), which renders them equivalent to bal-
≈ ·
0 Z sqq (s) 0 Ylqq (s) anced three-phase resistors at the line frequency. Within their
 
Z sdd (s) · Yldd (s) 0 voltage regulation bandwidth, they present CPL dynamics
= . (7)
0 Z sqq (s) · Ylqq (s) from a small-signal standpoint. This behavior is observed
in the dd element of their input impedance when aligning
As a diagonal matrix, the eigenvalues of L(s) are then simply the d-axis of the dq-coordinate with the active power flow
given by its diagonal components of the converter. The diagonal dominance of the AFE input
λ1 (s) = Z sdd (s)/Z ldd (s) (8) impedance matrix without EMI input filter has already been
reported in [8]; the corresponding input-admittance matrix
λ2 (s) = Z sqq (s)/Z lqq (s). (9)
YdqAFE is marked in Fig. 2. Mao et al. [8] also proposed a
This implies that the stability at ac interfaces can be assessed reduced-order model for the AFE converter, which, however,
directly from the impedance return ratios seen across the neglected the dynamics of the q-axis entirely. A similar
dd and qq interfaces. This is the exact dual of the stability approach was later proposed in [29], being limited by the
assessment in dc systems. The SISO gain margin (GM), same assumptions. More recently, Burgos et al. [30] also
phase margin (PM), and forbidden region concepts can be proposed the use of the scalar return ratio seen across the
readily applied to the dd and qq impedance ratios in (8) dd interface to assess the stability of high PF rectifiers fed
and (9). DC impedance specifications can then be applied by synchronous generators. Burgos et al. [30] also verified
separately to the dd and qq subsystems in question, resulting the scalar stability criterion in the presence of EMI filters
in much less conservative designs for the integration of power and reactive power loads; however, the EMI filter influence
electronics-based ac DPSs. In addition, most of the three-phase on the total input admittance matrix YdqLCAFE as marked
converters are controlled using dq-coordinate controllers. in Fig. 2 was not discussed. Last, the impact of the q-axis
1458 IEEE JOURNAL OF EMERGING AND SELECTED TOPICS IN POWER ELECTRONICS, VOL. 5, NO. 4, DECEMBER 2017

dynamics has also been recently studied showing how the

qq input impedance element of the AFE converter is shaped
by the PLL in the low-frequency range and beyond that by the
PFC current loop [15], [16]. This is especially critical under
regeneration or GTIs as instabilities could be easily triggered
by improperly tuned PLLs as presented in [18] and [19].
The input-admittance matrix of the AFE converter is derived
as (10). The reader is referred to [16] and [19] for a detailed
description
YdqAFE
 −1
I + Gid (I + Gdel Gci Gcv KGvd )−1 Gdel
=
·(−Gdei + Gci )K
 −1  Fig. 3. AFE input admittance when Iqs = 0 A.
Zin + Gid (I + Gdel Gci Gcv KGvd )−1 Gdel 
· d K .
· (Gdei − Gci )GPLL
i K − Gci Gcv KGve + GPLL
(10)
In (10), the converter dynamics are modeled by the MIMO
transfer functions. Specifically, Gvd corresponds to the transfer
function from duty cycle d̃AFE
s
to output dc voltage ṽ dcAFE
s
;

Gve is the transfer function of input ac voltage ṽ to output
s

dc voltage ṽ dcAFE
s ; Gid is the transfer function from d̃AFE s

to the boost inductor current ĩ s ; Z is the open-loop input

in
impedance. It is the function of the converter duty cycles [16],
and hence, changes in the operating point will modify its
dynamic response.
To capture the PLL dynamics two additional transfer
functions are introduced in the model, namely, GPLL d and
i
GPLL [16]. The first one corresponds to the transfer function
from ṽs to d̃s , and the second one corresponds to the transfer
AFE
function from ṽs to ĩ c [16]. It should be noticed that GPLL
d Fig. 4. AFE input admittance. (a)Iqs = −90% Ids (PF = 0.73), inductive
 reactive power. (b) Iqs = 90%Ids (PF = 0.73), capacitive reactive power.
is a function of the AFE steady-state duty cycle ( DAFE ), and
i
GPLL is a function of the steady-boost inductor current. Last,
the feedback controller transfer function Gci , the decoupling frequency response of Yqq [15]–[19], it is recommended then
transfer function Gdei [29], the signal conditioning filter K, that impedance specifications be issued for both d- and q-axes
and the delay transfer function Gdel —capturing the time in the dq-domain.
delay Tdel introduced by sampling and PWM—complete the Diagonal dominance could also be lost in the dq-domain
small-signal model of the AFE converter. impedance matrix, however, if the AFE converter operates
The input admittance is accordingly also a function of with nonunity PF—in reactive power compensation mode for
the operating point since it depends on GPLL d i
and GPLL . instance. As depending on the relative magnitude between
Further, it follows that when the converter operates with Iqs and Ids , GPLL
i could be nondiagonal, making YdqAFE
unity PF (Iqs = 0, DqAFE s ≈ 0), these two transfer nondiagonal too as shown in Fig. 4. However, in most DPSs
functions are diagonally dominant, which when combined where all the loads are interfaced with AFEs, systems will
with the effect of the feedback decoupling controller always work under nearly unity PF condition. No reactive
render YdqAFE diagonally dominant too. This is depicted compensation is needed. Therefore, the nondiagonal dominant
in Fig. 3, where a magnitude difference greater than 27 dB cases of YdqAFE can be ignored for DPS stability analysis.
between the diagonal (Ydd , Yqq ) and off-diagonal (Ydq , Yqd ) The impact of an LC filter on the AFE converter input
terms is observed. Although sufficient for practical purposes, admittance is assessed next, assuming a single-stage LC filter
this nonideal decoupling is the result of the digital delay and as shown in Fig. 5. In the case, multistage filters are in
parametric mismatch effect on the control system. The inherent question; these can be simplified to a single-stage equivalent
AFE converter characteristics are also observed in Fig. 3. for simplicity [36]. Accordingly, the AFE converter input
Specifically, Ydd reflects the CPL dynamics as it behaves as admittance YdqLCAFE , including the LC filter, is given by
a negative incremental resistor, being primarily shaped by YdqLCAFE = (ZdqLf + (YdqCf + YdqAFE )−1 )−1 (11)
the output voltage controller [8]. Yqq , on the other hand,
behaves as a positive incremental resistor as a result of the where ZdqLf and YdqCf correspond to the filter inductor
PLL synchronization [16]. Since instability could be trig- and filter capacitor dq-domain impedance and admittance,
gered by both the negative resistor behavior of Ydd and the respectively. There are certain design constraints that delimit
WEN et al.: AC STABILITY ANALYSIS AND dq FRAME IMPEDANCE SPECIFICATIONS 1459

Fig. 5. Simplified circuit schematic of AFE converter with LC input filter.

the boundary for the values of L f and C f . One is the

tolerable reactive power consumption by C f , typically 2–5%
of the active power of the converter [35]. With this condition,
C f is determined by
 2
Vds ωC f = 5%Vds Ids (12) Fig. 6. Ydq of an LC filter (a) without damping and (b) with damping.
5%
Cf = (13)
ωR The total input impedance is then given by the series connec-
where R is the base impedance of the AFE converter tion of ZdqLf and ZdqCAFE defined as follows:
system ZdqLCAFE
 
Vds = ZdqLf + ZdqCAFE =
Z dd Z dq
R= . (14) Z qd Z qq
Ids ⎡ ⎤
20Rω (s + 20ω)
The design of L f depends on the desired harmonic attenua- ⎢ + L f s −L f ω ⎥
= ⎣ s − 399ω
2 2
tion; regardless of the latter, its line frequency impedance is 20Rω (s − 20ω) ⎦. (19)
Lfω + L f s
typically 10% of R. s 2 − 399ω2
In order to simplify the following analysis, three fre- In D1 , L f s can be ignored, since L f s < L f ω. Then
quency ranges are defined spanning from dc to half of the        
switching frequency ( f sw /2), namely, range D1 from dc to the  Z dd  Z dd  20Rω(s +20ω) 1   400R 
 = =
line frequency ( f line ); range D2 from f line to the control band-  Z  Z   s 2 −399ω2 · L ω ≈ −399L ω (20)
 dq   qd   f
  f

width frequency ( f bw ); and range D3 from fbw to f sw /2 below  Z qq   Z qq  20Rω(s −20ω) 1   400R 
 = = · ≈ . (21)
which the average model is still valid. The input impedance  Z   Z   s 2 −399ω2 L f ω 399L f ω
qd dq
of AFE converters can, based on the preceding discussion,
be assumed to be diagonally dominant in D1 and D2 , since if Further, since the inductor impedances are typically less
the admittance is diagonally dominant, then its inverse matrix, than 10% of the load at the line frequency, the ratio of diagonal
the impedance matrix, must also be. In these frequency ranges, to off-diagonal impedance elements can be approximated by
 
ZdqAFE can be defined as  Z dd 
 
   Z  > 10 (22)
dq
−R 0
ZdqAFE = (15) which shows that ZdqLCAFE is diagonally dominant in D1 .
0 R
For D2 , L f s > L f ω, which renders (19) diagonally dominant
It follows, per Fig. 5, that: by simple inspection. Accordingly, since ZdqLCAFE is diago-
  nally dominant, then YdqLCAFE is too.
Z dd Z dq
ZdqCAFE = ((ZdqCf )−1 + (ZdqAFE )−1 )−1 = Beyond the roll-off frequency of the control loops, in D3 ,
Z qd Z qq
⎡ ⎤ the input admittance of the AFE converter is determined by the
20Rω (s + 20ω) 20Rω2 admittance of the LC filter. This matrix is nondiagonal at the
⎢ s 2 − 399ω2 s 2 − 399ω2 ⎥
=⎢⎣
⎥. (16) resonant frequency as shown in Fig. 6(a); however, if a proper
20Rω 2 20Rω (s − 20ω) ⎦ damping element is used, then this matrix becomes diagonally
− 2
s − 399ω2 s 2 − 399ω2 dominant around the resonant frequency as shown in Fig. 6(b).
Fig. 6 depicts the case where a series resistor of 0.5% of
The resultant impedance matrix is diagonally dominant
the capacitor impedance at the line frequency is added to the
according to (17) and (18)
capacitor; alternatively, active damping loops could be used to
       
 Z dd   Z dd   20Rω(s +20ω)   s + 20ω  achieve the same effect.
 = = = 
 Z dq   Z qd   20Rω2   ω  ≥ 20 (17) This result is verified in Fig. 7 that depicts the input-
        admittance magnitude of the AFE converter with an LC filter
 Z qq   Z qq   20Rω(s −20ω)   s − 20ω 
 = = = 
Z  Z   20Rω2   ω  ≥ 20. (18) under unity PF, evincing the diagonal dominance of this
dq qd dq-domain quantity.
1460 IEEE JOURNAL OF EMERGING AND SELECTED TOPICS IN POWER ELECTRONICS, VOL. 5, NO. 4, DECEMBER 2017

Fig. 7. AFE input admittance with an input filter and unit PF. Fig. 9. Z dd and Z dq of VSI with feedback decoupling control.

Fig. 8. Power stage and control system of VSI. Fig. 10. GTI.

B. Voltage Source Inverter

The VSI that is shown in Fig. 8 represents the typical
voltage source or power supply found in a DPS. These
converters are critical since if their output impedance is not
low enough, with certain loads the whole DPS could become
unstable [12]. To study its dynamics the small-signal model
of the VSI can be derived in the same fashion as that of
the AFE. The main difference is that the VSI has no PLL
and no dc bus voltage controller, as the inverter is assumed to
be fed from a stiff dc source, whether from batteries or another
source interfaced by a dc–dc converter, neglecting the impact
of any dc side dynamics. Instead, the VSI has a control loop to
regulate its ac output voltage in the dq domain. The resultant Fig. 11. GTI impedance.
small-signal model is linear and independent of the operating
point [37], [38]. parametric mismatches. Near the filter resonant frequency,
If, as desirable, a decoupling dq-coordinate controller and these terms are 15 dB less than the diagonal terms, showing
proper damping are used to control the VSI output volt- that the VSI output impedance ZdqVSI is diagonally domi-
age, its output impedance will also be diagonally dominant. nant. Furthermore, the linear nature of the VSI renders its
Specifically, the closed-loop control itself will help reduce impedance not influenced by its operating point.
the coupling between the dq-coordinate axes by effectively
rejecting the current and voltage crosstalk between channels. C. Grid-Tied Inverter
The use of additional decoupling loops will further reduce the The GTI—or regenerative rectifier—is also an important
cross-coupling effect [32], [33]. Damping, on the other hand, converter in the DPS. Fig. 10 shows the power stage and
will aid by attenuating the filter response around its resonant control system of this converter in the dq-coordinate, operating
frequency [36], [39]. in current-source mode from an ac bus standpoint. As seen,
As shown in [12] and [38], the Z dd and Z qq ele- both its power stage and ac control system are similar to those
ments of the VSI output impedance are identical; likewise, of the AFE converter. However, since the GTI is fed from a dc
Z dq and Z qd have the same magnitude but an 180° phase source, then its dc dynamics are neglected similar to that of
difference. Accordingly, Fig. 9 only shows Z dd and Z dq for the VSI. Intuitively, the output impedance of the GTI, namely,
a VSI with voltage and current feedback decoupling control. ZdqGTI , can also be designed to be diagonally dominant using
As evinced, at low frequency, the off-diagonal elements are decoupled dq-coordinate closed-loop controls [18].
greatly reduced by the decoupling terms in the current and Fig. 11 shows the output impedance of this converter in
voltage control loops, even under the effect of digital delay and dq domain. In Fig. 11, Z dd clearly reflects the current-source
WEN et al.: AC STABILITY ANALYSIS AND dq FRAME IMPEDANCE SPECIFICATIONS 1461

TABLE I
PARAMETERS OF E XPERIMENTAL T EST B ED

Fig. 12. Sample ac system with VSI feeding AFE converter.

Fig. 13. Test-bed hardware prototype for VSI feeding AFE system.

behavior in the d-axis, which is shaped in the low-frequency

range by the integrator of the current controller and by the
inductor impedance in the high-frequency range. Z qq , on
the other hand, behaves as a negative incremental resistor in
the low-frequency range, being dominated by the PLL dynam-
ics [18]. The off-diagonal terms Z dq and Z qd are very small
under unity PF as shown in Fig. 11 due to current feedback and
the decoupling controller [18]. Under nonunity PF, however, of the VSI and the forbidden regions of the impedance return-
which would be expected when operating in generic grid ratios seen across the dd and qq interfaces is demonstrated.
applications, the diagonal dominance is lost as shown in [18]. First, the impedances of the VSI are designed accord-
It follows then that the GTI can easily induce instability due to ing to the discussion in Section III-B. Fig. 14(a) shows
the negative incremental resistor behavior that characterizes its the measured output impedances of the VSI, clearly depict-
Z qq impedance element. Also, that when operating with unity ing its diagonal dominance as predicted. Then, a forbidden
PF—as expected within the DPS, stability in the system can region [5] on the polar plot of return ratios, Z ddVSI/Z ddAFE
be ensured by using impedance specifications in both dq frame and Z qqVSI /Z qqAFE , is defined using 6-dB GM and 60° PM,
axes. see Fig. 14(b). Based on Z ddVSI (the same as Z qqVSI ) and
the forbidden region, the specification for the impedances
IV. S IMPLIFIED S TABILITY A NALYSIS (Z ddAFE, Z qqAFE ) of the AFE can be illustrated in Fig. 14(c).
Specifically, admittance matrix of the AFE should be
A. Oscillation Caused by Constant Power Load designed diagonal dominantly. magnitudes of the diagonal
Fig. 12 shows the block diagram of an ac system under elements are at least 20 dB bigger than those of the one of
consideration, comprised of a VSI feeding an AFE converter. the off-diagonal elements. Then, the magnitudes of Z ddAFE
This configuration embodies the main power flow dynamics and Z qqAFE should be designed higher than those of the
encountered within a power-electronics-based DPS, where the “gain limit” in Fig. 14(c); in the frequency regions where
AFE converter is a CPL. In this case, the ac interface of the “gain limit” are broke, the phases of Z ddAFE and Z qqAFE
interest is the bus regulated by the VSI, representing the point should be smaller than those of the “upper phase limit” and
of common coupling where loads can connect. Fig. 13 shows bigger than those of the “lower phase limit.”
the corresponding test bed built to study this system, whose Fig. 15 shows the measured admittances of the AFE,
parameters are listed in Table I in the Appendix. depicting its diagonal dominance. Z ddAFE and Z qqAFE are
Experimental tests were conducted to verify the stability then plotted with their specifications as shown in Fig. 16.
assessment capability under decoupled dq-coordinate dynam- Z qqAFE is within the limits as shown in Fig. 16(b). Most
ics in the power-electronics-based DPS. They assessed the parts of Z ddAFE are within the limits. However, the phase of
stability at the ac interface in question with the AFE converter Z ddAFE is smaller than that of the “lower phase limit” from
operating with an LC input filter. The procedure of specifying 200 to 500 Hz, see Fig. 16(a). The VSI-AFE system is short
the input impedances of the AFE according to the impedances of PM from 200 to 500 Hz, yet still stable. An oscillation with
1462 IEEE JOURNAL OF EMERGING AND SELECTED TOPICS IN POWER ELECTRONICS, VOL. 5, NO. 4, DECEMBER 2017

Fig. 16. (a) Z ddAFE and (b) Z qqAFE with impedance specifications.

Fig. 14. (a) Measured ZdqVSI . (b) Return ratios and the forbidden region.
(c) Load impedance specification based on Z ddVSI and the forbidden region.

Fig. 17. Time-domain test of VSI feeding AFE with 320-V output voltage.

stable operating condition with a sustainable oscillation

of 319 Hz. This result validates the analysis from the Bode
plot using the impedance specifications.
Fig. 18 shows the corresponding characteristic loci of
the resultant return-ratio matrix L(s), superimposed with the
SISO impedance return ratios seen across the d- and q-axes
interfaces. Fig. 18 illustrates how Z ddVSI ·YddAFE and Z qqVSI ·
YqqAFE perfectly overlap the trajectories of λ1 and λ2 ,
verifying that stability at the ac interface can be fully assessed
Fig. 15. Measured AFE input admittances with 320-V output. by the direct inspection of the SISO return ratios avoiding the
use of the GNC.
a frequency between 200 and 500 Hz is likely to happen due Based on the aforementioned analysis, suggestions for the
to disturbances in the system. designer of the AFE can be given to shape the Z ddAFE of the
Fig. 17 shows the startup transient of the AFE when conn- AFE within the impedance specification. Fig. 19(a) shows the
ected to the VSI-controlled ac bus, illustrating a marginally modified Z ddAFE (with the smaller values of kpvAFE = 0.0628
WEN et al.: AC STABILITY ANALYSIS AND dq FRAME IMPEDANCE SPECIFICATIONS 1463

Fig. 20. AC system with both GTI and AFE.

Fig. 18. Eigenvalue loci of L(s) overplotted with diagonal elements ratios.

Fig. 21. Eigenvalue loci of L(s) over plotted with diagonal elements
ratios (a) stable case and (b) unstable case.

within a power-electronics-based DPS, stability can be

accurately assessed by using the SISO impedance return ratios
seen across the q-axis interface.
Accordingly, two experiments were conducted in this case
using the same test bed described and shown in Fig. 13, using
an additional ac power supply as the third converter to generate
and regulate the ac bus. The time-domain experimental results
can be found in [19, Fig. 27].
Using the models that are described in Section III and the
parameters listed in [19], impedances of ZdqGTI and Zdqgeq
at the ac interface in Fig. 20 are calculated. Using ZdqGTI
and Zdqgeq , Fig. 21 finally shows the characteristic loci at the
ac interface, where the trajectories of the eigenvalues of L(s)
Fig. 19. (a) Z ddAFE with slow dc voltage control. (b) Time-domain test overlap perfectly with the trajectories described by the SISO
results of VSI feeding AFE with slow dc voltage control.
impedance return ratios seen at the d- and q-axis interfaces,
see Fig. 21(a) and (b). Accordingly, the impedance across the
d- and q-axes can be used to assess stability treating the ac bus
and kivAFE = 3.2725);now, it is within the limits. The time-
as two independent dc interfaces. The impedance specification
domain test results shown in Fig. 19(b) also indicate a stable
procedure that is demonstrated in the last section can also be
operating condition of the system.
applied here for Z ddGTI and Z qqGTI . It will not be repeated
here for simplicity.
B. Grid-Synchronization Instability Caused by PLL
The instability induced by interactions of q-axis result- V. C ONCLUSION
ing from the synchronization of AFE and GTI convert- This paper has shown that under decoupled dq frame
ers in microgrid-type distribution systems as that shown dynamics—a characteristic of power-electronics-based DPSs
in [19, Fig. 20]. This is a unique type of interaction without using AFE converters, VSIs and GTIs to process power
any dual in dc systems. However, similar to other GNC-based flow, the system impedances seen at ac interfaces feature a
stability methods, the proposed approach relied on the diagonally dominant structure in the dq frame. As a result,
eigenvalues of the return-ratio matrix L(s) to evaluate sta- stability at these interfaces can be determined by the SISO
bility, veiling any direct relationship between the system impedance return ratios seen across the dq frame axes, namely,
dq-coordinate impedances and the stability of the system. Source ·Y Load and Z Source ·Y Load , as the trajectory described by
Z dd dd qq qq
The goal of this paper is accordingly to show that under them in the complex plane matches the trajectory, or charac-
decoupled dq-coordinate conditions, as would be expected teristic loci, of the eigenvalues of the return-ratio matrix L(s),
1464 IEEE JOURNAL OF EMERGING AND SELECTED TOPICS IN POWER ELECTRONICS, VOL. 5, NO. 4, DECEMBER 2017

itself product of the dq frame impedance and admittance [16] B. Wen, D. Boroyevich, P. Mattavelli, Z. Shen, and R. Burgos, “Influence
matrices ZSource
dq and YddLoad , which ultimately determine the of phase-locked loop on input admittance of three-phase voltage-source
converters,” in Proc. 28th Annu. IEEE Appl. Power Electron. Conf.
stability at the boundary per GNC. The relevance of this result Expo., Mar. 2013, pp. 897–904.
is threefold: first, it enabled the direct use of impedances [17] K. M. Alawasa, Y. A.-R. I. Mohamed, and W. Xu, “Active mitigation of
to assess stability in power electronics-based ac DPSs, thus subsynchronous interactions between PWM voltage-source converters
and power networks,” IEEE Trans. Power Electron., vol. 29, no. 1,
recovering the physical insight of the original impedance- pp. 121–134, Jan. 2014.
based analysis developed for dc systems; second, it enabled [18] B. Wen, D. Boroyevich, R. Burgos, P. Mattavelli, and Z. Shen, “Analysis
the use of straightforward impedance specification criteria to of D-Q small-signal impedance of grid-tied inverters,” IEEE Trans.
Power Electron., vol. 31, no. 1, pp. 675–687, Jan. 2016.
aid in the design and integration of these systems, all the while
[19] B. Wen, D. Dong, D. Boroyevich, P. Mattavelli, R. Burgos, and Z. Shen,
discarding the conservativeness of existent design criteria; and “Impedance-based analysis of grid-synchronization stability for three-
third, it identified the prominence of the dynamic interac- phase paralleled converters,” IEEE Trans. Power Electron., vol. 31, no. 1,
tions resulting from the synchronization of power converters, pp. 26–38, Jan. 2016.
[20] N. Bottrell, M. Prodanovic, and T. C. Green, “Dynamic stability of a
formulating the proposed method in both dq frame axes. microgrid with an active load,” IEEE Trans. Power Electron., vol. 28,
Experimental results with a laboratory scale power-electronics- no. 11, pp. 5107–5119, Nov. 2013.
based distributed power system were used to validate the [21] R. D. Middlebrook, “Measurement of loop gain in feedback systems,”
Int. J. Electron., vol. 38, no. 4, pp. 485–512, 1975.
approach.
[22] Y. Panov and M. M. Jovanovic, “Stability and dynamic performance of
current-sharing control for paralleled voltage regulator modules,” IEEE
R EFERENCES Trans. Power Electron., vol. 17, no. 2, pp. 172–179, Mar. 2002.
[23] J. Morroni, R. Zane, and D. Maksimovic, “An online stability margin
[1] P. W. Wheeler, J. C. Clare, A. Trentin, and S. Bozhko, “An overview of monitor for digitally controlled switched-mode power supplies,” IEEE
the more electrical aircraft,” Proc. Inst. Mech. Eng., G, J. Aerosp. Eng., Trans. Power Electron., vol. 24, no. 11, pp. 2639–2648, Nov. 2009.
vol. 227, no. 4, pp. 578–585, Apr. 2013. [24] J. Zeng, Z. Zhe, and W. Qiao, “An interconnection and damping
[2] C. D. Xu and K. W. E. Cheng, “A survey of distributed power system— assignment passivity-based controller for a DC–DC boost converter
AC versus DC distributed power system,” in Proc. Int. Conf. Power with a constant power load,” IEEE Trans. Ind. Appl., vol. 50, no. 4,
Electron. Syst. Appl., Jun. 2011, pp. 1–12. pp. 2314–2322, Jul./Aug. 2014.
[3] N. O. Sokal, “System oscillations from negative input resistance at [25] Y. Gu, W. Li, and X. He, “Passivity-based control of DC microgrid for
power input port of switching-mode regulator, amplifier, DC/DC con- self-disciplined stabilization,” IEEE Trans. Power Syst., vol. 30, no. 5,
verter, or DC/DC inverter,” in Proc. IEEE Power Electron. Spec. Conf., pp. 2623–2632, Sep. 2015.
Sep. 1973, pp. 138–140. [26] X. Zhang, X. Ruan, and Q.-C. Zhong, “Improving the stability of
[4] R. D. Middlebrook, “Input filter considerations in design and application cascaded DC/DC converter systems via shaping the input impedance
of switching regulators,” in Proc. IEEE Ind. Appl. Soc. Conf., Oct. 1976, of the load converter with a parallel or series virtual impedance,” IEEE
pp. 94–107. Trans. Ind. Electron., vol. 62, no. 12, pp. 7499–7512, Dec. 2015.
[5] C. M. Wildrick, F. C. Lee, B. H. Cho, and B. Choi, “A method of [27] M. K. Bakhshizadeh et al., “Couplings in phase domain impedance
defining the load impedance specification for a stable distributed power modeling of grid-connected converters,” IEEE Trans. Power Electron.,
system,” IEEE Trans. Power Electron., vol. 10, no. 3, pp. 280–285, vol. 31, no. 10, pp. 6792–6796, Oct. 2016.
May 1995. [28] A. Rygg, M. Molinas, C. Zhang, and X. Cai, “A modified sequence-
[6] S. D. Sudhoff, S. F. Glover, P. T. Lamm, D. H. Schmucker, and domain impedance definition and its equivalence to the dq-domain
D. E. Delisle, “Admittance space stability analysis of power elec- impedance definition for the stability analysis of AC power electronic
tronic systems,” IEEE Trans. Aerosp. Electron. Syst., vol. 36, no. 3, systems,” IEEE J. Emerg. Sel. Topics Power Electron., vol. 4, no. 4,
pp. 965–973, Jul. 2000. pp. 1383–1396, Dec. 2016.
[7] S. Vesti, T. Suntio, J. A. Oliver, R. Prieto, and J. A. Cobos, “Impedance- [29] B. Yin, R. Oruganti, S. K. Panda, and A. K. S. Bhat, “A simple single-
based stability and transient-performance assessment applying max- input–single-output (SISO) model for a three-phase PWM rectifier,”
imum peak criteria,” IEEE Trans. Power Electron., vol. 28, no. 5, IEEE Trans. Power Electron., vol. 24, no. 3, pp. 620–631, Mar. 2009.
pp. 2099–2104, May 2013. [30] R. Burgos, D. Boroyevich, F. Wang, K. Karimi, and G. Francis, “On the
[8] H. Mao, D. Boroyevich, and F. C. Y. Lee, “Novel reduced-order small- AC stability of high power factor three-phase rectifiers,” in Proc. IEEE
signal model of a three-phase PWM rectifier and its application in Energy Convers. Congr. Expo., Sep. 2010, pp. 2047–2054.
control design and system analysis,” IEEE Trans. Power Electron.,
[31] B. Wen, D. Boroyevich, R. Burgos, P. Mattavelli, and Z. Shen,
vol. 13, no. 3, pp. 511–521, May 1998.
“D-Q impedance specification for balanced three-phase AC distribution
[9] M. Belkhayat, “Stability criteria for ac power systems with regulated
power system,” in Proc. 30th Annu. Appl. Power Electron. Conf. Expo.,
loads,” Ph.D. dissertation, Purdue Univ., West Lafayette, IN, USA,
Mar. 2015, pp. 2757–2771.
Dec. 1997.
[10] J. Sun, “Small-signal methods for ac distributed power systems—A [32] D. Ke, L. Ping, K. Yong, and C. Jian, “Decoupling current control for
review,” IEEE Trans. Power Electron., vol. 24, no. 11, pp. 2545–2554, voltage source converter in synchronous roating frame,” in Proc. 4th
Nov. 2009. IEEE Int. Conf. Power Electron. Drive Syst., Oct. 2001, pp. 39–43.
[11] M. Schweizer and J. W. Kolar, “Shifting input filter resonances—An [33] B. Bahrani, S. Kenzelmann, and A. Rufer, “Multivariable-PI-based
intelligent converter behavior for maintaining system stability,” in Proc. dq current control of voltage source converters with superior axis
Int. Power Electron. Conf. (IPEC), May 2010, pp. 906–913. decoupling capability,” IEEE Trans. Ind. Electron., vol. 58, no. 7,
[12] B. Wen, D. Boroyevich, R. Burgos, P. Mattavelli, and Z. Shen, “Small- pp. 3016–3026, Jul. 2011.
signal stability analysis of three-phase AC systems in the presence [34] S.-K. Chung, “A phase tracking system for three phase utility interface
of constant power loads based on measured D-Q frame impedances,” inverters,” IEEE Trans. Power Electron., vol. 15, no. 3, pp. 431–438,
IEEE Trans. Power Electron., vol. 30, no. 10, pp. 5952–5963, May 2000.
Oct. 2015. [35] R. Teodorescu, M. Liserre, and P. Rodriguez, Grid Converters for
[13] J. Sun, “Impedance-based stability criterion for grid-connected invert- Photovoltaic and Wind Power Systems. Piscataway, NJ, USA: Wiley,
ers,” IEEE Trans. Power Electron., vol. 26, no. 11, pp. 3075–3078, 2011.
Nov. 2011. [36] N. Hensgens, M. Silva, J. A. Oliver, J. A. Cobos, S. Skibin, and
[14] X. Wang, F. Blaabjerg, and W. Wu, “Modeling and analysis of harmonic A. Ecklebe, “Optimal design of AC EMI filters with damping networks
stability in an AC power-electronics-based power system,” IEEE Trans. and effect on the system power factor,” in Proc. Energy Convers. Congr.
Power Electron., vol. 29, no. 12, pp. 6421–6432, Dec. 2014. Expo., Sep. 2012, pp. 637–644.
[15] L. Harnefors, M. Bongiorno, and S. Lundberg, “Input-admittance cal- [37] S. Hiti, D. Boroyevich, and C. Cuadros, “Small-signal modeling and
culation and shaping for controlled voltage-source converters,” IEEE control of three-phase PWM converters,” in Proc. IEEE Ind. Appl. Soc.
Trans. Ind. Electron., vol. 54, no. 6, pp. 3323–3334, Dec. 2007. Annu. Meet., Oct. 1994, pp. 1143–1150.
WEN et al.: AC STABILITY ANALYSIS AND dq FRAME IMPEDANCE SPECIFICATIONS 1465

[38] B. Wen, D. Boroyevich, R. Burgos, and P. Mattavelli, “Modeling the Dushan Boroyevich (S’81–M’86–SM’03–F’06)
output impedance of three-phase uninterruptible power supply in D-Q received the Dipl. Ing. degree from the University of
frame,” in Proc. Energy Convers. Congr. Expo., Sep. 2014, pp. 163–169. Belgrade, Belgrade, Serbia, in 1976, the M.S. degree
[39] K. Hatua, A. K. Jain, D. Banerjee, and V. T. Ranganathan, “Active from the University of Novi Sad, Novi Sad, Serbia,
damping of output LC filter resonance for vector-controlled VSI- in 1982, in what then used to be Yugoslavia, and the
fed AC motor drives,” IEEE Trans. Ind. Electron., vol. 59, no. 1, Ph.D. degree from the Virginia Polytechnic Institute
pp. 334–342, Jan. 2012. and State University (Virginia Tech), Blacksburg,
VA, USA, in 1986.
From 1986 to 1990, he was an Assistant Profes-
sor and the Director of the Power and Industrial
Electronics Research Program with the Institute for
Power and Electronic Engineering, University of Novi Sad, and then became
an acting Head of the Institute. He then joined the Bradley Department
of Electrical and Computer Engineering, Virginia Tech, as an Associate
Professor. He is currently the American Electric Power Professor with the
Bradley Department of Electrical and Computer Engineering and the Co-
Bo Wen (S’11–M’15) received the B.S. degree Director of the Center for Power Electronics Systems, Virginia Tech. His
in electrical engineering from Xi’an Jiaotong Uni- current research interests include multiphase power conversion, electronic
versity, Xi’an, China, in 2006, and the M.S. and power distribution systems, power electronics systems modeling and control,
Ph.D. degrees in electrical engineering from Virginia and integrated design of power converters.
Tech, Blacksburg, VA, USA, in 2011 and 2014, Prof. Boroyevich was the President of the IEEE Power Electronics Society
respectively. for 2011–2012. He was a member of the U.S. National Academy of Engi-
He was a Research Associate at the Department neering. He was a recipient of the IEEE William E. Newell Power Electronics
of Engineering, Electrical Engineering Division, Technical Field Award.
University of Cambridge, Cambridge, U.K. He is
currently a Lecturer with the School of Electri- Paolo Mattavelli (S’95–A’96–M’00–SM’10–F’14)
cal and Electronic Engineering, the University of received the M.S. (with Hons.) and the Ph.D.
Manchester, Manchester, U.K. His current research interests include multi- degrees in electrical engineering from the Univer-
phase power conversion, power electronics systems modeling, and stability sity of Padua, Padua, Italy, in 1992 and 1995,
analysis. respectively.
From 1995 to 2001, he was a Researcher at the
University of Padua. From 2001 to 2005, he was
an Associate Professor at the University of Udine,
Udine, Italy, where he led the Power Electronics
Laboratory. In 2005, he joined the University of
Padua, Vicenza, Italy. From 2010 to 2012, he was
a Professor and a member of the Center for Power Electronics Systems at
Virginia Tech, Blacksburg, VA, USA. He is currently a Professor with the
Rolando Burgos (S’96–M’03) received the B.S. University of Padua. His current google scholar h-index is 60. His current
degree in electronics engineering, the Electronics research interest includes analysis, modeling and analog and digital control of
Engineering Professional degree, and the M.S. and power converters, grid-connected converters for renewable energy systems and
Ph.D. degrees in electrical engineering from the Uni- microgrids, and high-temperature and high-power density power electronics.
versity of Concepción, Concepción, Chile, in 1995, In these research fields, he has been leading several industrial and government
1997, 1999, and 2002, respectively. projects.
In 2002, he joined the Center for Power Elec- From 2003 to 2012, he served as an Associate Editor of the IEEE
tronics Systems (CPES), Virginia Tech, Blacksburg, T RANSACTIONS ON P OWER E LECTRONICS . From 2005 to 2010, he was
VA, USA, as a Post-Doctoral Fellow, where he then the Industrial Power Converter Committee Technical Review Chair of the
became a Research Scientist in 2003 and a Research IEEE T RANSACTIONS ON I NDUSTRY A PPLICATIONS. For terms 2003–2006,
Assistant Professor in 2005. In 2009, he joined ABB 2006–2009, and 2013–2015, he has been a member-at-large of the IEEE Power
Corporate Research, Raleigh, NC, USA, as a Scientist and then became Electronics Society’s Administrative Committee. He was a recipient of the
a Principal Scientist in 2010. In 2010, he was appointed as an Adjunct Prize Paper Award in the IEEE T RANSACTIONS ON P OWER E LECTRONICS
Associate Professor of the Electrical and Computer Engineering Department in 2005, 2006, 2011 and 2012, and the 2nd Prize Paper Award in 2007 at the
at North Carolina State University, Raleigh, NC, USA, where he was with the IEEE Industry Application Annual Meeting.
Future Renewable Electric Energy Delivery and Management Systems Center.
In 2012, he joined Virginia Tech, where he is currently an Associate Professor Zhiyu Shen (S’11–M’13) received the B.S. and
with the Bradley Department of Electrical and Computer Engineering and M.S. degrees in electrical engineering from Tsinghua
a Faculty Member with the CPES. His current research interests include University, Beijing, China, and the Ph.D. degree
multiphase multilevel power conversion, grid power electronics systems, in electrical engineering from Virginia Tech,
stability of ac and dc power systems, high power density power electronics, Blacksburg, VA, USA, in 2004, 2007, and 2013,
modeling, and control theory and applications. respectively.
Dr. Burgos is a member of the IEEE Power Electronics Society, where He was a Research Scientist with the Center for
he currently serves as an Associate Editor of the IEEE T RANSACTIONS Power Electronics System, Virginia Tech. In 2016,
ON P OWER E LECTRONICS , the IEEE P OWER E LECTRONICS L ETTERS , and he joined General Electric Global Research,
the IEEE J OURNAL OF E MERGING AND S ELECTED T OPICS IN P OWER Niskayuna, NY USA, as a Power Electronics Engi-
E LECTRONICS . He is also a member of the IEEE Industry Applications neer. His current research interests include three-
Society and the IEEE Industrial Electronics Society. He is the Vice Chair of phase ac system impedance measurement, three-phase ac system small-signal
the Power and Control Core Technologies Committee of the Power Electronics stability, high-frequency high-density converter design, and control system
Society. architecture in high-power converters.