2078
4, APRIL 2012
Generalized Average Modeling of Dual Active
Bridge DC–DC Converter
Hengsi Qin, Student Member, IEEE, and Jonathan W. Kimball, Senior Member, IEEE
Abstract—Full-order continuous-time average modeling and dy-
namic analysis of bidirectional dc–dc dual active bridge (DAB)
converters are studied. The transformer current in DAB converter
is purely ac, making continuous-time modeling difficult. The pro-
posed full-order continuous-time average model uses the dc terms
and first order terms of transformer current and capacitor volt-
age as state variables, resulting in a third-order model, if capac-
itor equivalent series resistance (ESR) is not considered, and a
sixth-order model if ESR is considered. A control-to-output-voltage
transfer function is derived for DAB converters. Experimental re-
sults confirm that the proposed model correctly predicts the small-
signal frequency response and an even more accurate prediction
can be obtained if capacitor ESR is taken into account.
Index Terms—Average modeling, bidirectional dc-dc converter,
dual active bridge converter, phase shift modulation.
I. INTRODUCTION
HE dual active bridge (DAB) converter was originally
T proposed as an alternative to the high-power dc–dc con-
verter [1]. It is an isolated, bidirectional, buck and boost dc–dc
converter topology, which can be used in high-power appli-
cations for solid-state transformers [2], [3], smart grids [4],
electric/hybrid vehicles [5], as well as interface among three dc
ports [6]. The advantages of DAB converters over other can-
didate topologies include a lower number of passive compo-
nents, high power density, and high power efficiency resulting
from zero voltage switching (ZVS). Recent advancements in dc–
dc DAB converters include design optimization [7], a detailed
model to evaluate power loss [5], PWM control schemes [8], ad-
vanced modulation schemes [9], advanced switching strategies
to reduce loss [10]–[12], and detailed models to address some
parasitic nonlinear effects [13], [14].
A closed-loop controller is needed when the output of a DAB
converter needs to be regulated. The design of the controller
requires a small-signal average model of the DAB converter.
There are two approaches to modeling a DAB converter: 1) A
simplified reduced-order model neglects the transformer cur-
rent dynamic [15]–[17]; and 2) a full-order discrete-time model
preserves the dynamic of transformer current [18], [19]. The
discrete-time model is one approach to modeling those convert-
ers with large variation and resonant operations [20]. A full-
Manuscript received March 20, 2011; revised June 8, 2011 and July 19,
2011; accepted August 3, 2011. Date of current version February 20, 2012.
Recommended for publication by Associate Editor T. Suntio.
The authors are with the Department of Electrical and Computer Engineering,
Missouri University of Science and Technology, Rolla, MO 65409 USA (e-mail:
hengsi.qin@mail.mst.edu; kimballjw@mst.edu).
Digital Object Identifier 10.1109/TPEL.2011.2165734
0885-8993/$26.00 © 2011 IEEE
IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 27, NO.
Fig. 1. DAB converter schematic.
order discrete-time model can obtain higher accuracy at low
frequencies than a reduced-order model can [19]. However, a
continuous-time model is usually preferred because it provides
more physical insight and facilitates control design.
The conventional averaging technique for dc–dc converters
requires negligible current ripple [21]. This is not satisfied in
DAB converters because transformer current is purely ac. In-
stead, a generalized averaging technique, which uses more terms
in the Fourier series of state variables, is used in the present
work [22], [23]. Generalized average modeling has been applied
to single-phase ac–dc rectifier [24] and dc–ac resonant inverter
system [25]. A full-order continuous-time state-space average
model for a dc–dc DAB converter is developed. As compared to
the previous reduced-order continuous-time model [15]–[17],
the proposed model is more accurate at dc and low frequency,
and also properly captures the effects of capacitor ESR.
This paper is organized as follows: Section II describes the
operation principles of DAB dc–dc converters as background.
Section III presents the development of a full-order continuous-
time average model of DAB converters using the general-
ized averaging approach. Section IV discusses the small-signal
control-to-output transfer function based on the proposed aver-
age model. Section V discusses the effect of capacitor equiva-
lent series resistance. Simulation and experimental results are
provided in Section VI. Finally, Section VII summarizes the
contributions of this work.
II. DAB DC–DC CONVERTERS
The schematic of a DAB dc–dc converter is given in Fig. 1. It
consists of two H-bridges and a high-frequency transformer. The
high-frequency transformer provides both galvanic isolation and
energy storage through winding leakage inductance. There is a
resonant capacitor in parallel with each transistor-diode pair in
order to enable ZVS.
A dc–dc DAB converter can be controlled by 1) the phase shift
φ between two H-bridges; (2) the duty ratios of switching de-
vices; and (3) the switching frequency. For simplicity, this work
only analyzes phase shift modulation (PSM) at fixed switching
frequency and fixed 50% duty ratio. Power is transferred from
QIN AND KIMBALL: GENERALIZED AVERAGE MODELING OF DUAL ACTIVE BRIDGE DC–DC CONVERTER 2079

Fig. 3. Simplified DAB converter schematic.

and the switching function at output bridge, s2 (τ ), is

⎧
Fig. 2. Waveforms during one switching cycle.
⎪ dT T dT
⎨ 1, ≤τ < +
s2 (τ ) = 2 2 2 (5)
the leading bridge to the lagging bridge, which is given by ⎪
⎩ −1, 0 ≤ τ < dT T dT
or + ≤ τ < T.
vs vo 2 2 2
P = d (1 − d) (1) If we select transformer current, it , and output capacitor voltage,
2N fs Lt
vo , as state variables, then the state equations of a DAB converter
where N is the turns ratio of the high-frequency transformer, vs
can be derived as
is the input voltage, vo is the output voltage, phase shift ratio d
is equal to πφ , φ is the phase shift in radians, fs is the switching dit (τ ) Rt s1 (τ ) s2 (τ )
= − it (τ ) + vi (τ ) − vo (τ ) (6)
frequency, and Lt = NL 12 + L2 is the leakage inductance of the dτ Lt Lt Lt
high-frequency transformer, referred to the output side. The in- and
herent symmetry of circuit topology in DAB converters ensures
bidirectional power flow. Fig. 2 depicts the typical transformer dvo (τ ) 1 s2 (τ ) IN
=− vo (τ ) + it (τ ) − . (7)
current and input/output voltages waveforms during one switch- dτ RCo Co Co
ing period.
Equations (6) and (7) are both time-varying and nonlinear.
III. GENERALIZED MODEL OF DAB CONVERTERS To derive a linear time-invariant model, the next step is to apply
The model is derived by assuming that transformer magnetiz- averaging. The basic idea behind averaging is to represent a
ing current is not significant, the voltage drop across transistor- state variable, x(τ ), during the interval t − T ≤ τ ≤ t, using its
diode pairs is negligible, and transistor switching transients are Fourier series
negligible. The input capacitance Cin is usually relatively large. ∞

Therefore, the dynamics of the input capacitor are not consid- x(τ ) = xk (t)ej k ω s τ (8)
ered in this work (Fig. 3). The current source on the output may k =−∞

be of either polarity for bidirectional power flow. The voltage at

where ωs = 2πfs , and complex number xk (t) is the k th co-
the input side is referred to the output side, which is represented
efficient in the Fourier series
as Vi = VNs . Transformer leakage inductance at both windings 
is lumped as an equivalent inductance Lt , referred to the output 1 t
xk (t) = x(τ )e−j k ω s τ dτ.
side. Similarly, transistor on-time resistance and transformer T t−T
winding resistance are lumped as an equivalent resistance Rt 
(Fig. 3). Using PSM, the voltage at the transformer input side, 1 t
= x(τ ) cos kωs τ dτ
vp , can only have two states: 1) +Vi when transistors Q1 and T t−T
Q2 are ON; and 2) −Vi when transistors Q3 and Q4 are ON. 
1 t
Therefore, −j x(τ ) sin kωs τ dτ. (9)
T t−T
vp (τ ) = s1 (τ )vi (τ ) (2)
This formulation is equivalent to that discussed in [22], where
where the switching function at input bridge, s1 (τ ), is τ = t − T + s and s ∈ [0, T ).
⎧ T Conventional state-space average can be applied when the
⎨ 1, 0≤τ < ripples in state variables are small, which, in fact, only considers
s1 (τ ) = 2 (3)
⎩ T dc terms (k = 0) in (8) and (9). However, the transformer current
−1, ≤τ <T in a DAB converter is purely ac, meaning ripple is large and the
2
dc term is zero. The generalized averaging technique uses more
where T = 1/fs . Throughout the analysis, τ is used to denote
terms in the Fourier series to represent more details in the model.
the time within a period, referenced to the rising edge of s1 .
In the case of DAB converters, it is natural to include more terms
Similarly, at output bridge
(both k = 0 and k = ±1) in the Fourier series to represent the
vs (τ ) = s2 (τ )vo (τ ) (4) averages of state variables.
2080 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 27, NO. 4, APRIL 2012

According to [22], [23], the derivative of the kth coefficient dvo 1R 1 1
for variable x is = − iN 1R − vo 1R + ωs vo 1I
dt Co RCo

d d 1 1
xk (t) = x (t) − jkωs xk (t) (10) + s2 0 it 1R + s2 1R it 0 (19)
dt dt k Co Co
where  dt
d
xk represents the average of the differential of a state dvo 1I 1 1
= − iN 1I − vo 1I − ωs vo 1R
variable in (6) and (7). The kth coefficient of the product of two dt Co RCo
variables x and y is 1 1
+ s2 0 it 1I + s2 1I it 0 . (20)
∞
 Co Co
xyk = xk −i yi . (11) It is assumed that the dynamics of the input voltage and the
i=−∞
load is much slower that those of the DAB converter. Therefore,
Consider that the 1st coefficient and the −1st coefficient in vi 0 = Vi , vi 1R = vi 1I = 0, iN 0 = iN , and iN 1R =
Fourier series are complex conjugates, the product of zeroth iN 1I = 0.
coefficient terms is Next, it is necessary to find the coefficients of switching
functions s1 (t) and s2 (t). Because of the fixed 50% duty ratio,
xy0 = x0 y0 + 2 (x1R y1R  + x1I y1I ) (12) the zeroth coefficients of s1 (t) and s2 (t) are both zero
and the 1st coefficient terms are s1 0 = s2 0 = 0. (21)
xy1R = x0 y1R + x1R y0 (13) The 1st coefficients of s1 (t) and s2 (t) are
xy1I = x0 y1I + x1I y0 (14) s1 1R = 0 (22)
where the subscripts “R” and “I” mean the real and the imagi- 2
s1 1I = − (23)
nary parts of a complex number, respectively. π
Applying (10) and (12)–(14) to the state-space equations in and
(6) and (7), the zeroth and 1st coefficients of state variables it
2 sin dπ
and vo are s2 1R = − (24)
π
dit 0 Rt 1 2
= − it 0 + s1 0 vi 0 + s1 1R vi 1R s2 1I =−
2 cos dπ
. (25)
dt Lt Lt Lt π
2 1
+ s1 1I vi 1I − s2 0 vo 0
Lt Lt Finally, substitute (21)–(25) into (15)–(20), and orga-
2 2 nize differential equations into matrix representation in (26)
+ s2 1R vo 1R + s2 1I vo 1I (15) (shown in the bottom of the next page), –> where X =
Lt Lt
[vo0 it1R it1I it0 vo1R vo1I ]T . In (26), it0 , vo1R , and vo1I are
dit 1R Rt all zero if their initial values are zero. The dynamics of it0 ,
= − it 1R + ωs it 1I
dt Lt vo1R , and vo1I are decoupled from the rest of system. There-
1 fore, the state-space representation in (26) can be equivalently
+ (s1 0 vi 1R + s1 1R vi 0 ) reduced to
Lt
⎡ ⎤
1 1 4 sin (dπ) 4 cos (dπ)
− (s2 0 vo 1R + s2 1R vo 0 ) (16) ⎡ ⎤ ⎢ − − −
RCo πCo πCo ⎥
d ⎣ o0 ⎦ ⎢ ⎥
Lt v
⎢ 2 sin (dπ) Rt ⎥
dit 1I Rt it1R = ⎢ − ωs ⎥
= − it 1I − ωs it 1R dt ⎢ πL L ⎥
dt Lt it1I ⎣ 2 cos (dπ)
t t
Rt ⎦
−ωs −
1 πLt Lt
+ (s1 0 vi 1I + s1 1I vi 0 ) ⎡ ⎤
Lt ⎡ ⎤ 1
vo0 0 −  
1 ⎢ Co ⎥ vi
− (s2 0 vo 1I + s2 1I vo 0 ) (17) ⎣ ⎦
× it1R + ⎣ 0 ⎢ 0 ⎦⎥ . (27)
Lt 2 iN
it1I − 0
and πLt
dvo 0 1 1 The state equation in (27) indicates that the dynamics of DAB
= − iN 0 − vo 0 dc–dc converters can be represented using the zeroth coefficient
dt Co RCo
of output voltage and the 1st coefficients of transformer current
1 2 as state variables. If more terms in the Fourier series were used,
+ s2 0 it 0 + s2 1R it 1R
Co Co then the model would be more accurate. However, the resulting
2 model would be too complex to provide insightful information
+ s2 1I it 1I (18) for controller design.
Co
QIN AND KIMBALL: GENERALIZED AVERAGE MODELING OF DUAL ACTIVE BRIDGE DC–DC CONVERTER 2081

⎡ ⎤
IV. SMALL-SIGNAL AVERAGE MODEL Δvo0
Controller design and stability analysis for a power converter × ⎣ Δit1R ⎦
require the derivation of small-signal control-to-output transfer Δit1I
⎡ 4 ⎤
C o (I0I sin (πD) − I0R cos (πD))
function, which is the dynamic response of a converter from
a small perturbation in the control signal. In (27), when there +⎣ 2V o 0 ⎦ Δd.
L t cos (πD)
is a small perturbation in d, the state variables vo0 , it1R , and
it1I will deviate from their steady states. Assume that the input − L t sin (πD)
2V o 0

voltage is constant, and define the deviations of state variables (34)

as
Substituting (28)–(33) into (27), the small-signal model of the
Δd = d − D (28) DAB converter is given in (34).–>
Δvo0 = vo0 − Vo0 (29)
V. EFFECT OF CAPACITOR EQUIVALENT SERIES RESISTANCE
Δit1R = it1R − It1R (30)
The average model in the previous sections does not consider
Δit1I = it1I − It1I (31) the effect of output capacitor equivalent series resistance (ESR).
where capital variables represent steady states, lower case vari- This section discusses the influence of ESR on the generalized
ables mean large-signal states, and Δ variables mean small- average model of DAB converters.
signal states. When ESR is considered, the state space equations of DAB
Equation (27) contains multiplication of control input and converters in (6) and (7) are revised as
state variables. For small Δd, the nonlinear term sin (πd) vo0 is dit (t) Rt s2 Rc R s1 (t)
approximated by using (28)–(31) = − it (t) − + vi (t)
dt Lt Lt Rc + R Lt
sin (πd) vo0 = sin (πD) Δvo0 + Vo0 sin (πD) cos (πΔd) s2 (t) R
− vo (t) (35)
+ Vo0 cos (πD) sin (πΔd) Lt Rc + R
dvc (t) 1 s2 (t) R
= sin (πD) Δvo0 + Vo0 sin (πD) =− vo (t) + it (t)
dt (R + Rc ) Co Co Rc + R
+ Vo0 cos (πD) (πΔd) . (32)
1
− IN (t) (36)
Other nonlinear terms are approximated in a similar way. Co
Steady state values of vo0 , it1R , and it1I can be found from
and
(27) by solving
⎡ ⎤ R Rc R Rc R
v vo (t) = vc (t) + s2 it (t) − IN (t)
d ⎣ o0 ⎦ Rc + R Rc + R Rc + R
it1R = 0. (33) (37)
dt
it1I in which capacitor voltage vc is different from output voltage
⎡ ⎤ vo in this case.
1 4 sin (Dπ) 4 cos (Dπ)
⎡ ⎤ ⎢ − − − In (35), (36), and (37), the same generalized averaging tech-
RCo πCo πCo ⎥
Δvo0 ⎢ ⎥ nique described in (12)–(14) can also be applied. Therefore, the
d ⎣ ⎢ 2 sin (Dπ) R ⎥
Δit1R ⎦ = ⎢ −
t
ωs ⎥ 0-index and 1-index of state variables are given in (38) (shown
dt ⎢ πLt Lt ⎥
Δit1I ⎣ 2 cos (Dπ) ⎦ in the bottom), where K1 = R cR+R and K2 = RRc c+R R
. –> Com-
−ωs −RLt
t
paring (38) and (26), it is obvious that the dynamics of all six
πLt
⎡ ⎤
1 4 sin (dπ) 4 cos (dπ)
− − − 0 0 0
⎢ RC πCo πCo ⎥
⎢ o ⎥ ⎡ ⎤
⎢ 2 sin (dπ) Rt ⎥ 1
⎢ − ωs 0 0 0 ⎥ 0 −
⎢ πLt Lt ⎥ ⎢ ⎥
⎢ ⎥ ⎢
Co
⎥
⎢ 2 cos (dπ) Rt ⎥ ⎢ 0 0 ⎥ 
⎢ −ωs − 0 0 0 ⎥ ⎢ 2 ⎥ Vi
d ⎢ ⎥ −
X=⎢ πLt Lt X +⎢ 0 ⎥
dt ⎢ Rt 4 sin (dπ) 4 cos (dπ) ⎥
⎥ ⎢ πL
⎢ 0 t
⎥ IN .
⎥
⎢ 0 0 0 − ⎥ ⎢ 0 ⎥
⎢ Lt πLt πLt ⎥ ⎣ 0 ⎦
⎢ ⎥ 0
⎢ 2 sin (dπ) 1 ⎥
⎢ 0 0 0 − − ωs ⎥ 0 0
⎢ πCo RCo ⎥
⎣ 2 cos (dπ) 1 ⎦
0 0 0 − −ωs −
πCo RCo
(26)
2082 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 27, NO. 4, APRIL 2012

TABLE I
CIRCUIT PARAMETERS

state variables are now coupled. As a result, it is impossible to

reduce (38) into a third-order system.
Small-signal perturbation can be applied to (38) as well to
derive a control-to-output transfer function. The procedure is
the same as previously discussed and it obtains similar results.
Therefore, it is not repeated here.
VI. SIMULATION AND EXPERIMENT RESULTS
Simulation and experiment results are provided in order to
verify the proposed average model. A time-domain switching
model has been developed in PLECS. An experiment setup
has been built. The main parameters for the power stage of
the simulation model and experimental prototype are given in
Table I. Switching devices are IGBT STGF19NC60HD from
STMicroelectronics, rated at 19 A, 600 V. The switching
frequency of IGBTs is 10 kHz. A high-frequency trans-
former is made up of a UU-100 ferrite core from Ferrox-
cube. Both primary and secondary windings consist of 40
turns and each turn is made up of four AWG 18 mag-
netic wires. Series inductors are added on both windings to
form the necessary winding inductance for the proper oper-
ation of the DAB converter prototype. Two 350 V 270 μF
aluminium electrolytic capacitors are connected in parallel
to form the output capacitor. A 4.7 nF ceramic capacitors
is connected in parallel with each IGBT to reduce switch-
ing loss. This DAB converter prototype is controlled using a
32-bit floating-point microprocessor TMS320F28335 from
Texas Instruments.
The control-to-output transfer function is a metric for small-
signal average model of power converters. Three averaging tech-
niques are compared:
1) Reduced-order continuous-time model derived from
steady-state transformer current [15], [16].
Fig. 4. Diagram to measure control-to-output transfer function.
2) Full-order discrete-time model [18], [19].
3) The proposed full-order continuous-time model.
The test conditions are: The input voltage is 100 V, and the
voltage transfer ratio is set to 1; the load resistance at the output
side is set to 38 Ω, and output power is 260 W. The control-to-
output transfer function of a DAB dc–dc converter was measured
using a varying frequency sinusoidal signal superimposed on a
steady-state duty ratio. The procedure to measure control-to-
output transfer function is described as follows (see Fig. 4). In
controller software, a function generator generated a sinusoidal
perturbation signal at frequency fper t , which is specified as an
input. The perturbation signal was superimposed on steady-state
duty ratio. The superimposed duty ratio was fed into the on-chip
PWM generator to control the switching devices in the power
stage of DAB converter. The magnitude of perturbation signal,
di,per t , was set to 0.01 when fper t was less than 512 Hz and 0.02
when fper t was larger than 512 Hz. The output voltage of DAB
converter was measured and recorded using a Tektronix MSO-
4054 oscilloscope. Measured data were imported and processed
in MATLAB. The magnitude of the output voltage perturba-
tion, vper t , at fo,per t was extracted by FFT analysis. Therefore,
control-to-output gain at fper t is given by
 
vo,per t
Gain = 20 log10 (39)
di,per t
where di,per t = 0.01 when fper t < 512 Hz and di,per t = 0.02
when fper t > 512 Hz.

⎡ ⎤
− (R +R1 c )C o − 4K 1πsin(dπ
Co
)
− 4K 1 πcos(dπ
Co
)
0 0 0 ⎡ ⎤
⎢ 2K 1 sin(dπ ) 2K 2 sin(dπ ) ⎥ 0 − C1o
⎢ −R t
ωs 0 0 ⎥
⎢ 2K πcos(dπ
Lt Lt π Lt ⎥ ⎢ 0 0 ⎥ 
⎢ 1 )
−ωs −R 2K 2 cos(dπ ) ⎥ ⎢ ⎥
⎢ 0 0 ⎥ ⎢− 2 ⎥ Vi
t
d π Lt Lt π Lt 0
X= ⎢ ⎥ X + ⎢ π Lt ⎥ .
dt ⎢ 0 4K 2 sin(dπ ) 4K 2 cos(dπ )
−R t 4K 1 sin(dπ ) 4K 1 cos(dπ )
⎥ ⎢ 0 0 ⎥ IN
⎢ π Lt π Lt Lt π Lt π Lt ⎥ ⎣ ⎦
⎢ 2K 1 sin(dπ )
− πCo − (R +R1 c )C o ⎥ 0 0
⎣ 0 0 0 ωs ⎦
0 0
0 0 0 − 2K 1 πcos(dπ
Co
)
−ωs − (R +R1 c )C o
(38)
QIN AND KIMBALL: GENERALIZED AVERAGE MODELING OF DUAL ACTIVE BRIDGE DC–DC CONVERTER 2083

The effect of ESR becomes significant at the frequency where

1/ωCo is equal to Rc , which is 1180 Hz for the capacitors used
in this experimental prototype. Therefore, Fig. 5(a) shows that
the difference between the experimental measurement and the
average model becomes larger above this frequency. Fig. 5(b)
shows the models with ESR considered. The reduced-order av-
erage model fails to predict the control-to-output gain at high
frequency. The higher-order average model in (38) provides
more accurate match than the third-order model in (26) does.
Fig. 5(b) also indicates a tradeoff between model complexity
and model accuracy. A simpler, third-order model is preferred if
the effect of ESR can be ignored, while a complex, higher order
model is necessary in the case that ESR must be considered.

VII. CONCLUSION
A novel average modeling approach for a phase-shift con-
trolled bidirectional dc–dc DAB converter is presented in this
paper. The proposed method, which uses switching frequency
terms in the Fourier series of state variables, captures the effects
of purely ac transformer current on converter dynamics. The pro-
posed model is compared with the previously presented model-
ing methods. Both simulation and experimental results show that
the proposed model provides more accurate frequency-domain
control-to-output gain. In particular, the proposed model is more
accurate at dc and low frequency than the reduced-order models
in [15]–[17]. The proposed average model for DAB converter
is in continuous-time domain and is insightful and friendly to
controller design.
The effect of capacitor ESR is considered when the zeroth and
1st coefficients of both transformer current and output capacitor
voltage are kept. A higher-order model is able to obtain more
accurate predictions at the cost of complexity.

Fig. 5. Calculated and measured control-to-output gain. Proposed full-
order continuous-time model (dashed line). Reduced-order continuous-time
model (solid line). Full-order discrete-time model (dash-dot line). Re-
sults from detailed switching simulation (crosses). Results from hardware
measurement (circles). All gains are measured in dB. (a) Without ESR.
(b) With ESR.
The small-signal control-to-output gains of the three previ-
ously mentioned models are plotted in Fig. 5. The obtained
results from the PLECS simulation and the hardware experi-
ment show a very good match between calculated and measured
transfer functions. Fig. 5(a) shows the models without consid-
ering ESR. There are some mismatches between the reduced-
order continuous-time model and the measured results at low
frequencies. On the other hand, both the discrete-time model
and the proposed full-order continuous-time model provides a
more accurate results than does the reduced-order model at low
frequency. The proposed approach obtains similar accuracy as
the discrete-time model does. However, the proposed approach
provides a continuous-time model, which is easy to manipulate
and more convenient for controller design.
REFERENCES
[1] M. N. Kheraluwala, R. W. Gascoigne, D. M. Divan, and E. D. Baumann,
“Performance characterization of a high-power dual active bridge,” IEEE
Trans. Ind. Appl., vol. 28, no. 6, pp. 1294–1301, 1992.
[2] S. Bhattacharya, T. Zhao, G. Wang, S. Dutta, S. Baek, Y. Du, B. Parkhideh,
X. Zhou, and A. Q. Huang, “Design and development of generation-i
silicon based solid state transformer,” in Proc. 25th Annu. IEEE Appl.
Power Electron. Conf. Expo., Palm Springs, CA, 2010, pp. 1666–1673.
[3] J. Shi, W. Gou, H. Yuan, T. Zhao, and A. Huang, “Research on voltage
and power balance control for cascaded modular solid-state transformer,”
IEEE Trans. Power Electron., vol. 26, no. 4, pp. 1154–1166, Apr. 2011.
[4] S. Inoue and H. Akagi, “A bidirectional isolated dc-dc converter as a core
circuit of the next-generation medium-voltage power conversion system,”
IEEE Trans Power Electron., vol. 22, no. 2, pp. 535–542, 2007.
[5] F. Krismer and J. Kolar, “Accurate power loss model derivation of a high-
current dual active bridge converter for an automotive application,” IEEE
Trans Ind. Electron., vol. 57, no. 3, pp. 881–891, Mar. 2010.
[6] C. Zhao, S. D. Round, and J. W. Kolar, “An isolated three-port bidirectional
dc–dc converter with decoupled power flow management,” IEEE Trans.
Power Electron., vol. 23, no. 5, pp. 2443–2453, 2008.
[7] C. Mi, H. Bai, C. Wang, and S. Gargies, “Operation, design and control
of dual h-bridge-based isolated bidirectional dc–dc converter,” Power
Electronics, IET, vol. 1, no. 4, pp. 507–517, 2008.
[8] A. Jain and R. Ayyanar, “PWM control of dual active bridge: Comprehen-
sive analysis and experimental verification,” IEEE Trans. Power Electron.,
vol. 26, no. 4, pp. 1215–1227, Apr. 2011.
[9] Z. Haihua and A. M. Khambadkone, “Hybrid modulation for dual-active-
bridge bidirectional converter with extended power range for ultracapac-
itor application,” IEEE Trans. Ind. Appl., vol. 45, no. 4, pp. 1434–1442,
2009.
2084
[10] H. Bai and C. Mi, “Eliminate reactive power and increase system efficiency
of isolated bidirectional dual-active-bridge dc-dc converters using novel
dual-phase-shift control,” IEEE Trans. Power Electron., vol. 23, no. 6,
pp. 2905–2914, Nov. 2008.
[11] G. G. Oggier, G. O. Garcia, and A. R. Oliva, “Switching control strategy
to minimize dual active bridge converter losses,” IEEE Trans. Power
Electron., vol. 24, no. 7, pp. 1826–1838, 2009.
[12] G. Oggier, G. Garci anda, and A. Oliva, “Modulation strategy to operate
the dual active bridge dc–dc converter under soft switching in the whole
operating range,” IEEE Trans. Power Electron., vol. 26, no. 4, pp. 1228–
1236, Apr. 2011.
[13] B. Hua, C. C. Mi, and S. Gargies, “The short-time-scale transient processes
in high-voltage and high-power isolated bidirectional dc-dc converters,”
IEEE Trans. Power Electron., vol. 23, no. 6, pp. 2648–2656, 2008.
[14] Y. Xie, J. Sun, and J. S. Freudenberg, “Power flow characterization of a
bidirectional galvanically isolated high-power dc/dc converter over a wide
operating range,” IEEE Trans. Power Electron., vol. 25, no. 1, pp. 54–66,
2010.
[15] H. K. Krishnamurthy and R. Ayyanar, “Building block converter module
for universal (ac–dc, dc–ac, dc–dc) fully modular power conversion archi-
tecture,” in Proc. IEEE Power Electron. Spec. Conf., 2007, pp. 483–489.
[16] H. Bai, M. Chunting, W. Chongwu, and S. Gargies, “The dynamic model
and hybrid phase-shift control of a dual-active-bridge converter,” in Proc.
34th Annu. Conf. IEEE Ind. Electron. (IECON), 2008, pp. 2840–2845.
[17] H. Bai, Z. Nie, and C. C. Mi, “Experimental comparison of traditional
phase-shift, dual-phase-shift, and model-based control of isolated bidirec-
tional dc-dc converters,” IEEE Trans. Power Electron., vol. 25, no. 6,
pp. 1444–1449, 2010.
[18] F. Krismer and J. W. Kolar, “Accurate small-signal model for the digital
control of an automotive bidirectional dual active bridge,” IEEE Trans.
Power Electron., vol. 24, no. 12, pp. 2756–2768, 2009.
[19] C. Zhao, S. D. Round, and J. W. Kolar, “Full-order averaging modelling of
zero-voltage-switching phase-shift bidirectional dc-dc converters,” IET
Power Electron., vol. 3, no. 3, pp. 400–410, 2010.
[20] G. C. Verghese, M. E. Elbuluk, and J. G. Kassakian, “A general approach
to sampled-data modeling for power electronic circuits,” IEEE Trans.
Power Electron., vol. PE-1, no. 2, pp. 76–89, 1986.
[21] R. W. Erickson and D. Maksimovic, Fundamentals of Power Electronics,
2nd ed. New York: Springer-Verlag, 2001.
[22] S. R. Sanders, J. M. Noworolski, X. Z. Liu, and G. C. Verghese, “Gen-
eralized averaging method for power conversion circuits,” IEEE Trans.
Power Electron., vol. 6, no. 2, pp. 251–259, 1991.
[23] V. A. Caliskan, O. C. Verghese, and A. M. Stankovic, “Multifrequency
averaging of dc/dc converters,” IEEE Trans. Power Electron., vol. 14,
no. 1, pp. 124–133, 1999.
IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 27, NO. 4, APRIL 2012
[24] C. Gaviria, E. Fossas, and R. Grino, “Robust controller for a full-bridge
rectifier using the IDA approach and GSSA modeling,” IEEE Trans.
Circuits Systems I: Regular Papers, vol. 52, no. 3, pp. 609–616, Mar.
2005.
[25] Z. Ye, P. Jain, and P. Sen, “Phasor-domain modeling of resonant inverters
for high-frequency ac power distribution systems,” IEEE Trans. Power
Electron., vol. 24, no. 4, pp. 911–923, Apr. 2009.
Hengsi Qin (S’09) received the B.S. and M.S. de-
gree in electrical engineering from Central South
University, Changsha, Hunan, China, in 2005 and
2008, respectively. He is currently working toward
the Ph.D. degree at the Department of Electrical and
Computer Engineering, Missouri University of Sci-
ence and Technology (Missouri S&T), Rolla.
His current research interests include the analysis,
design, control, and optimization of power electronic
converters in renewable electric power grid.
Jonathan W. Kimball (M’96–SM’05) received the
B.S. degree in electrical and computer engineering
from Carnegie Mellon University, Pittsburgh, PA, in
1994, and the M.S. degree in electrical engineering
and the Ph.D. degree in electrical and computer en-
gineering from the University of Illinois at Urbana-
Champaign, Urbana-Champaign, in 1996 and 2007,
respectively.
From 1996 to 1998, he was with Motorola,
Phoenix, AZ, where he was engaged in designing
insulated gate bipolar transistor (IGBT) modules for
industrial applications. He then joined Baldor Electric, Fort Smith, AR, where
he designed industrial adjustable-speed drives ranging 1–150 hp. In 2003, he
joined the University of Illinois as a Research Engineer, where he later became a
Senior Research Engineer. In 2003, he cofounded SmartSpark Energy Systems,
Inc., Champaign, IL, where he was Vice President of Engineering. In 2008, he
joined Missouri University of Science and Technology (Missouri S&T), Rolla,
MO, where he is currently an Assistant Professor.
Dr. Kimball is a member of Eta Kappa Nu, Tau Beta Pi, and Phi Kappa Phi.
He is a licensed Professional Engineer in the State of Illinois.