Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 27, NO.

2, FEBRUARY 2012 891

Asymmetrical Duty Cycle Control and Decoupled


Power Flow Design of a Three-port Bidirectional
DC-DC Converter for Fuel Cell Vehicle Application
Lei Wang, Member, IEEE, Zhan Wang, Student Member, IEEE, and Hui Li, Senior Member, IEEE

Abstract—This paper proposes a new asymmetrical duty cy- the Toyota FCHV (Fuel Cell Hybrid Vehicle) in December 2002,
cle control method for a three-port bidirectional DC-DC converter and Honda started to deliver fuel cell vehicles (the Honda FCX)
with two current-fed ports interfacing with low voltage battery and to clients in Japan and the United States in 2002. Fuel cell ve-
ultracapacitor in a fuel cell vehicle. Along with the phase shift con-
trol managing the power flow between the ports, asymmetric duty hicles (FCVs) are considered one of the next generation vehicle
cycle is applied to each port to maintain a constant DC bus voltage technologies.
at low voltage side, which as a result will achieve wide zero-voltage- Since fuel cells have slow dynamics and lack energy storage
switching (ZVS) range for each port under varied ultracapacitor capabilities, an energy storage system such as a battery pack is
and battery voltages. The ZVS range analysis of different duty cy- required for fuel cell powered vehicle applications to improve
cle control methods as well as the circulation power loss between
the ports have been analyzed. In addition, the power flow design the system dynamics and recoup the braking energy. However,
featuring the reduced coupling factors between the ports have been batteries have a substantial discharge power deficiency at low
developed for the three-port bidirectional DC-DC converter. A fuel temperature and limited current capability at high or low state-
cell vehicle power train including a 2.5 kW three-port DC-DC con- of-charge (SOC). On the other hand, ultracapacitors have high
verter interfacing a 12 V battery and ultracapacitor was built in power density, reciprocal charge-discharge performance, high
the laboratory. The proposed asymmetrical duty cycle control and
power flow design was implemented and verified on the hardware efficiency, and long calendar life. Therefore, active hybrid en-
test bed under urban driving cycle. The experimental results val- ergy storage including both battery and ultracapacitor technolo-
idated that proposed asymmetrical duty cycle method has higher gies, as well as a bidirectional DC-DC converter, has been pro-
efficiency than other methods; furthermore, they also validated the posed in [3], [14], [18], [19]. This solution can significantly
reduced coupling factor between phase shift control and duty cycle improve vehicle performance characteristics such as system ef-
control.
ficiency in cold weather, as well as extend the battery cycle and
Index Terms—Asymmetrical duty cycle control, fuel cell vehicle, calendar life [3]. Furthermore, fuel cell vehicles will also re-
power flow design, three-port bidirectional DC-DC converter. quire a bidirectional DC-DC converter to interconnect the fuel
cell powered high-voltage bus and the low-voltage buses for ve-
hicle auxiliary loads. One way to combine these functions is to
I. INTRODUCTION utilize the vehicle 12 V battery, a low voltage ultracapacitor with
INCE climate change and energy source shortages have be- a multiport bidirectional DC-DC converter, as shown in Fig. 1.
S come an urgent challenge, governments around the world
have put stricter regulations and constraints on emissions and
This bidirectional DC-DC converter is required to provide sev-
eral kW of power during the energy storage discharging mode
energy efficiency. Fuel cells use hydrogen as fuel to produce to boost the voltage of the high-voltage bus to around 300 V
electricity and their only byproduct is water; so they are basi- during start-up. It also can provide some peak power during
cally emission free. As a result, fuel cells have attracted increas- acceleration and store the energy captured by the regenerative
ing interest in vehicle applications. General Motor built a fuel braking for a short time period [4], [5], [9].
cell vehicle in 1967, and in recent years, a number of fuel cell One of the most popular concepts for bidirectional DC-
vehicles have been developed by companies and government DC converters is the dual-active-bridge topology [6]–[9], [15],
supported projects [1], [2]. Toyota began limited marketing of which utilizes either two full-bridges or two half-bridges at the
primary and secondary sides of a high frequency transformer
with phase shift control resulting in flexible power flow con-
trol and zero-voltage-switching (ZVS). A number of three-port
bidirectional DC-DC converters, which utilize this principle,
Manuscript received January 19, 2011; revised April 8, 2011; accepted June
7, 2011. Date of current version January 9, 2012. This work was supported by are reported in [10]–[14]. In [10]–[12], full-bridge voltage-fed
the National Science Foundation under Award Number ECCS-0641972. Rec- circuits are chosen for each port. To maintain ZVS conditions
ommended for publication by Associate Editor S. Choi. under wide input voltage range, the duty cycle has to be ad-
L. Wang is with GE Global Research, Niskayuna, NY 12309 USA.
Z. Wang and H. Li are with the Department of Electrical and Computer justed for both upper and lower switches, which would cause
Engineering, Florida A&M University-Florida State University College of En- high input current ripple. Half-bridge circuits are used in [13],
gineering, Tallahassee, FL 32310 USA (e-mail: hli@caps.fsu.edu). [14] as a comparison. In [13], one boost half-bridge circuit is
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. employed for the ultracapacitor port, while the other two ports
Digital Object Identifier 10.1109/TPEL.2011.2160405 are still voltage-fed half bridges. A unified duty cycle method

0885-8993/$26.00 © 2011 IEEE


892 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 27, NO. 2, FEBRUARY 2012

Fig. 1. Proposed fuel cell vehicle system configuration using three-port bidirectional DC-DC converter to interface 12 V battery, low voltage ultracapacitor, and
fuel cell powered high-voltage bus.

was presented in this paper to control all three ports; this method voltage range. The circulating power loss within this three-port
can be referred to as symmetrical duty cycle (SDC) control. Al- converter was also analyzed. It was found that minimal circu-
though the duty cycle modulation could extend ZVS for the lating loss is inherently guaranteed in this converter with phase
ultracapacitor port, the fuel cell port cannot maintain ZVS in shift operating principle, and the leakage inductance between
a wide range since its voltage varies. In [14], two current-fed each port can be equal or different according to the specific de-
half-bridge circuits are proposed for both the battery and ul- sign. Additionally, the steady state power transfer function and
tracapacitor port as an appropriate topology; however, the duty small signal model of this converter were developed. To reduce
cycle is fixed at 50% for all three ports, which can be referred the coupling factor between phase shift and duty cycle closed
to as fixed duty cycle (FDC). When the input voltage of the two loop, specially designed bandwidth for two control loops were
current-fed ports varies in a wide range, the DC bus voltage of proposed; furthermore, this design also addresses the issue of
every half-bridge also varies in a wide range since they are just right-half-plane (RHP) zero within the plant function.
doubled by the input voltage with FDC method. As a result, To verify the proposed asymmetrical duty cycle control and
ZVS capabilities will be lost. dedicated control design for this multivariable system, this three-
Although the full-bridge based bidirectional DC-DC con- port DC-DC converter was utilized to interface 12 V battery
verter may have less rms loss than its half-bridge counterparts, (BU) and low voltage ultracapacitor (UC), and a 300 V high-
the half-bridge topology is still advantageous for the several voltage bus in a fuel cell vehicle power train test bed built in
kilowatt power applications in this paper. First, a half-bridge the laboratory. The test bed configuration is shown in Fig. 1. It
topology has less device numbers, leading to a reduced number includes a 10 kW PM dynamometer with a flywheel and was
of gate drivers and auxiliary power supplies. Second, a full- tested under an urban driving cycle. The experimental results
bridge topology requires twice the voltage applied on the trans- of the system performance validate that the proposed control
former, leading to a doubled transformer winding number in could achieve full range ZVS, and thus a better efficiency than
order to get the same magnetic flux density. Since the DC resis- other control methods in a wide power range. Furthermore, the
tance and AC resistance will also increase when the turns num- experimental results also validate the reduced coupling factor
ber is doubled, the transformer rms loss of full-bridge topology between phase shift and duty cycle control in this system.
will not be significantly less than that of half-bridge topology.
Moreover, if the input voltage has a wide variation range such II. ASYMMETRICAL DUTY CYCLE CONTROL AND
as the ultracapacitor, the current-fed topology with duty cycle CIRCULATION POWER ANALYSIS
control for half-bridge is more appropriate than full-bridge. This
is because the full-bridge topology under duty cycle control has A. Topology Description
lower power transfer capability than a half-bridge topology due Triple-half-bridges (THB) bidirectional DC-DC converters
to a zero voltage level being applied on the transformer, which are considered as an appropriate choice in this paper to in-
does not exist in half-bridge circuit [12], [18]. terface the BU and UC. Fig. 2 shows the circuit topology of
Therefore, the triple half-bridge topology introduced in [14] this THB converter: which can be seen as the current-current-
with a proposed new asymmetrical duty cycle (ADC) control voltage (C-C-V) topology, i.e., two current-fed ports and one
was studied in this paper to interconnect a low voltage ultra- voltage-fed port. C-C-V topology was firstly reported in [14].
capacitor, a 12 V battery, and a high-voltage bus in the fuel Current-fed port has the following advantages over voltage-fed
cell vehicle application. The phase shifts were still used for the port when interfacing with energy storage element: 1) smaller
power-flow control, but the duty cycle of half bridge in every current ripple for the BU and UC at steady states; 2) boosted
port is different and adjustable based on its own input voltage. As DC voltage on the transformer primary side reduces the current
a result the ZVS can be maintained for all the ports at full input requirement for isolated transformer design which results in
WANG et al.: ASYMMETRICAL DUTY CYCLE CONTROL AND DECOUPLED POWER FLOW DESIGN 893

Fig. 2. Three-port Triple Half Bridge (THB) DC-DC converter with two current-fed ports interfacing with low voltage battery (BU) and ultracapacitor (UC).

improved transformer efficiency; and 3) the current mode con- to the variation of input DC voltages of the UC and BU. The
trol is available to achieve more control flexibility for this con- duty cycle of v2 is fixed at 50% since it is a voltage-fed port.
verter, as well as soft current slope for the BU. Therefore, the The power flow between three ports is shown in Fig. 4, where
UC can provide/absorb fast peak power during the load tran- P1 , P2 , and P3 are input power of each port, P12 , P23 , and
sients, and the BU life cycle can be increased by reducing its P31 are power transfer between each port. All the voltages and
current stress. The current mode control can also provide over impedances are referred to port 1 side. Following the calculation
current protection for the BU and UC. method in [9], they are derived as below in ADC method:

⎪ φ12 (π − |φ12 |) ∓ (0.5 − D1 )2 π 2
B. Phase Shift and Duty Cycle Control ⎪
⎪ P = V12 V34


12
con12


The power transfer between every two ports in the three-port ⎪

converter is independent with the third port, and it is the same φ23 (π − |φ23 |) ∓ (0.5 − D3 )2 π 2
P23 = V56 V34 CASE I

⎪ con
with the one derived from a two-port converter. Therefore, Pm k ⎪

23


can be derived in (1) based on methods of [6], [9], ⎪
⎩P31 = φ31 (π − |φ31 |) ∓ (D1 − D3 ) π V56 V12
2 2

φm k (π − |φm k |) con31
Pm k = Vm Vk (1) ⎧
πωLrm k ⎪P = 2 D1 πφ12 ∓ (0.5 − D1 )φ12 V V
2


where m, k(1, 2, 3), Pm k is the power between port m and port ⎪

12
con12
12 34


k, Lrm k is the leakage inductance between m and k port, Vm ⎪

and Vk are the magnitudes of transformer winding voltages as 2 D3 πφ23 ∓ (0.5 − D3 )φ223
P23 = V56 V34 CASE II (2)

⎪ con23
shown in Fig. 2, φm k is shown in Fig. 3. ⎪



On the other hand, the duty cycle of each half-bridge can ⎪
⎩P31 = 4 D1 D3 πφ31 ∓ (D1 − D3 )φ31 V56 V12
2

be adjusted to control half-bridge DC voltage. There are three con31
types of duty cycle control methods for this C-C-V topology:
Fixed Duty Cycle control (FDC), Symmetrical Duty Cycle con- where con12 = 4πωLr 12 , con23 = 4πωLr 23 , con31 =
trol (SDC), and Asymmetrical Duty Cycle control (ADC). The 4πωLr 31 . D1 and D3 are the varied duty cycles for port 1
former two are introduced in [13], [14], the last one is the pro- and port 3, while port 2 has fixed duty cycle as 0.5. The power
posed new control method. The idealized transformer voltage flow calculation is related to the duty cycle range, if duty cycle
waveforms of FDC and SDC are shown in Fig. 3(a) and (b), displacement is smaller than corresponding phase shift displace-
with those of ADC shown in Fig. 3(c). In all the three methods, ment in given port, it is calculated based on equations of CASE I,
the phase shift φ32 and φ12 are adjusted to control the power otherwise it will be derived using equations of CASE II.
flow. The duty cycle of v1 , v2 and v3 are fixed at 50% in FDC
control where v1 , v2 , and v3 are transformer voltages as shown C. Zero Voltage Switching (ZVS) Analysis
in Fig. 2. During SDC control, v1 , v2 , and v3 have the same Based on the transformer voltage v1 , v2 , and v3 waveforms
varied duty cycle D. For ADC control, v1 , and v3 has different in Fig. 3(c), the switches sequence can be derived as follows:
duty cycle D1 and D3 , whereas D1 and D3 change according S1 and S2 are switched at t3 and t0 , S3 and S4 are switched at
894 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 27, NO. 2, FEBRUARY 2012

Fig. 3. Idealized transformer voltage waveforms at different duty cycle control methods: (a) fixed 50% duty cycle control, (b) symmetrical duty cycle control,
and (c) asymmetrical duty cycle control.

t5 and t2 , and S5 and S6 are switched at t4 and t1 . The ZVS S5 : iL r 3 (t4 ) − iLdc3 (t4 ) < 0
is maintained at the condition that a positive current should
S3 : iL r 2 (t5 ) < 0 (3)
flow through the switch that is going to be turned off. For the
voltage-fed port, the currents of switches are only determined by
transformer currents, but for the current-fed port, the currents of where iLdc1 and iLdc3 are input currents for two current-fed
switches are also determined by input DC inductor current. The ports, iL r 1 , iL r 2 , and iL r 3 are transformer currents for every
ZVS conditions for all the switches S1–S6 are derived below: port. The switches and the currents of (3) can be seen in Fig. 2.
As shown in Fig. 3(c), the transformer current and input DC
inductor current waveform are depending on the phase shift and
S2 : iL r 1 (t0 ) − iLdc1 (t0 ) > 0
duty cycle, once the phase shift and duty cycle are given, both
S6 : iL r 3 (t0 ) − iLdc3 (t1 ) > 0 slopes of the currents can be calculated, then the current value
at all the switching moments can be derived. Since (3) shows
S4 : iL r 2 (t2 ) > 0
that the ZVS of S2 and S6 are the most difficult to maintain
S1 : iL r 1 (t3 ) − iLdc1 (t3 ) < 0 due to the DC current injection, the ZVS of S2 is selected and
WANG et al.: ASYMMETRICAL DUTY CYCLE CONTROL AND DECOUPLED POWER FLOW DESIGN 895

the highlighted red area is derived according to the practical


operation range of the converter. First, the duty cycle range is
from 0.3 to 0.5 due to the voltage of ultracapacitor will vary
from the rated value. Second, the phase-shift angle is limited
to (−π/4, π/4) to avoid serious reactive power loss [17]. When
the input voltages of the two current-fed ports are different,
adopting SDC method can cause the DC voltage of one of them
to differ from the other two, thus the transformer current slope is
not increasing at t0 , which can cause less ZVS margin and even
loss to ZVS. It is observed in Fig. 5 that ADC method has wider
range of ZVS than SDC method. Based on the same reason,
adopting FDC method will also make the three DC voltages on
the half-bridge different, thus the transformer current slope is
not increasing at t0 , which will decrease ZVS range too.
The above analysis is based on ideal conditions. If switch
capacitances are taken into consideration, they tend to decrease
the ZVS range since more current margin is needed to discharge
the capacitors; however, another parameter as switch dead time
makes the circuit has more time to discharge the capacitor before
Fig. 4. Three-port DC-DC converter power flow diagram. another switch to turn on, that will at least in part compensate
for the decrease of ZVS range caused by switch capacitances.

D. Circulation Power Loss Analysis


Both battery and ultracapacitor are low voltage devices in this
application. With such low input voltages minimizing rms loss
in the system is equally important to achieving a wide range of
ZVS. The proposed asymmetrical duty cycle control not only
can achieve a wide ZVS range, but also reduce peak current
and rms current resulting in lower rms loss. In addition, the
circulation current exists in the three-port bi-directional DC-DC
converter leading to extra rms loss; therefore, how to minimize
the circulation power loss is investigated in this section.
A power flow chart of a three-port phase shift converter sys-
tem is shown in Fig. 4. The power transferred from port 1 to port
2, from port 2 to port 3, and from port 3 to port 1, is denoted as
Fig. 5. ZVS range of S2 under worst case when ΔIL d c 1 = 0: (a) with sym-
P12 , P23 , and P31 , respectively. Furthermore, the fundamental
metrical duty cycle control and (b) with asymmetrical duty cycle control. rms values of the currents between the ports are also denoted
as I12 , I23 , and I31 . When phase shift angle between every port
is small, which is also the design requirement for phase shift
analyzed in detail in (4)–(6) converter, and also the transformer voltages are close to each
1 other, the power between every port are
iL r 1 (t0 ) > ILdc1 − ΔILdc1
avg (4)
2
P12 = U1 I12
where ILdc1 avg is the average DC inductor current
P23 = U2 I23
P1
ILdc1 avg = (5)
Vin1 P31 = U3 I31 (7)
and ΔILdc1 is the peak-to-peak value of the DC inductor current where U1 , U2 , and U3 represent the fundamental RMS values
(1 − D)Vin1 of transformer voltages of every port.
ΔILdc1 = . (6) The total loss between every port then can be calculated as
fs Ldc1
According to (4), the input current ripple ΔILdc1 is helpful to 2
Ploss = R12 I12 2
+ R23 I23 2
+ R31 I31 (8)
maintain ZVS. Since Ldc1 is large enough to reduce the current
ripple in this application, the worst case of ZVS for S2 is when where R12 , R23 , R31 is the equivalent resistance between every
ΔILdc1 = 0. The ZVS range of S2 under worst case when port and can be in Fig. 4.
ΔILdc1 = 0 can be derived from (4)–(6) and shown in Fig. 5(a) The target of reducing circulation energy within a multiport
and (b) for both the SDC and ADC methods. In Fig. 5(b), system is equivalent to the target of minimizing the conduction
896 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 27, NO. 2, FEBRUARY 2012

loss between the ports


min Ploss . (9)
I 1 2 ,I 2 3 ,I 3 1 ,

While P1 , P2 , and P3 are predefined by power management, the


target is to find the optimized power flow P12 , P23 , and P31 to
minimize system loss. Since
P12 = P1 + P31 , P23 = P2 + P12 , P31 = P3 + P23 .
(10)
The objective can be further derived to
    2  2
2
P12 P2 + P12 P12 − P1
min R12 + R23 + R31 .
P1 2 U1 U2 U3
(11)
By solving its derivative, the solution to achieve minimum
loss is Fig. 6. Proposed control system for THB converter.
R31 P1 − R23 P2
P12 = . (12)
R12 + R23 + R31
choosing big inductance between port 1 and port 3, since trans-
P23 and P31 can also be derived from (10), respectively. In former leakage inductance directly affects the power flow, such
the symmetrical case, Lr 1 = Lr 2 = Lr 3 , assume L and R are design will decrease the coupling between port 1 and port 3 thus
proportional, then R12 = R23 = R31 will be derived, it comes a smaller phase shift Φ32 at given power will be achieved.
to the solution
P1 − P2 III. TRIPLE HALF BRIDGE CONVERTER CONTROL
P12 =
3 (POWER FLOW) DESIGN
P12 + P23 + P31 = 0. (13) The developed control system is shown in Fig. 6. The UC
In another special case, if Lr 2 = 0, Lr 1 = Lr 3 then 2Lr 12 = and BU are connected to the inverter DC link through this three-
2Lr 23 = Lr 31 , 2R12 = 2R23 = R31 , it comes to port C-C-V THB converter. The control block consists of the
BU-port control and UC-port control where varied duty cycles
2P1 − P2
P12 = and phase shift angles are generated and sent to the converter.
4
In Fig. 6, port 1 is connected to ultracapacitor so the control
P1 − P3 variables of UC port are φ12 and D1 . φ12 and D1 can be seen
P12 + P23 + P31 = . (14)
4 in Fig. 3(c) where the former represents the phase shift angle
In addition, (15) is derived based on the converter operating between transformer voltage v1 and v2 , and the latter is the duty
principle and can also be observed in Fig. 3 cycle of port 1. Similarly, port 3 is connected to the battery and
the control variables of BU port are defined as φ32 and D3 based
φ12 + φ23 + φ31 = 0. (15)
on Fig. 3(c).
Since R is proportional to the inductance between each port, Fig. 7 shows the detailed control implementations of BU-port
and the power between each port is nearly proportional to the Φ control and UC-port control when they are connected to the THB
and inverse proportional to the L, as shown in (1) and (2), (16) converter. The THB converter is represented by its plant transfer
is derived below functions in Fig. 7. The control objectives are fuel cell averaged
φ12 φ23 φ31 power, battery and ultracapacitor SOC, V12 and V56 . V12 is the
P12 + P23 + P31 ∝ + + . (16) LVS (low voltage side) DC link voltage at UC port and V56 is
ωLr 12 ωLr 23 ωLr 31
the LVS (low voltage side) DC link voltage at BU port. They
Based on (16) it can be concluded that (13) and (14) are can be found in Fig. 2. The UC-port control and BU-port control
naturally fulfilled in phase shift operating principle. Therefore, are described below.
different inductances between each port can be chosen to satisfy
specific design, and it will not affect minimum loss principle. In
the special case when R12 = R23 = R31 , (13) is satisfied. A. UC-port Control
Although the fundamental power loss can be minimized by Fig. 6 shows that the two control variables of UC port are
optimum power flow, the peak transformer current in this phase φ12 and D1 . Fig. 7 shows that φ12 is used to control FC average
shift converter will affect system loss. Fig. 3(c) shows that peak power and BU state-of-charge (SOC), and D1 is used to control
transformer current is mostly dependent on the phase shift an- V12 , the LVS DC link voltage of UC port.
gle when the input power of each port is low. The ideal case The fuel cell average power control method is similar to the
is that the phase shift angle is zero when the input power of method presented in [19]–[22]. In this paper, it is implemented
corresponding port is zero, which requires the port 1 and port by controlling fuel cell average current Ifc to follow its reference
3 to be completely decoupled. Such decoupling is achieved by Ifc∗ . The Ifc∗ is generated considering multiple factors. The first
WANG et al.: ASYMMETRICAL DUTY CYCLE CONTROL AND DECOUPLED POWER FLOW DESIGN 897

Fig. 8 Coupled multiinput multioutput network and its controller.

with a red circle in Fig. 7. Moreover, the parameters in the


Fig. 7. Control schemes of UC port and BU port with power stage plant UC SOC controller are not constant and varied with different
functions.
ranges of VUC . Therefore, the UC can provide/absorb most of
the power at the beginning of load transients and then BU will
factor is related to the average load power, which can be derived take over to provide most of the power. G11 2 is the converter
approximately from Iload at selected driving cycle; the second plant function where input is φ32 and output is VUC . Similar to
is depends on the system Joule loss (I2 R) information that can port 1, V56 is maintained by duty cycle of port 3 (D3 ) through a
be estimated from IBU and IUC . If the above estimations have cascaded voltage and current control loop where G22A 2 is the
deviations, the BU SOC will deviate from its reference value. plant function interfacing with D3 and output IBU , and G22B 2
Therefore, Ifc∗ can be further adjusted by controlling the BU is the plant function between IBU and V56 .
SOC to the desired value in a long run at very slow sampling
rate. G11 represents the converter plant function that connected C. Design With Reduced Coupling Factor
to Ifc control loop. All the control is implemented in digital As shown in Fig. 7, this THB converter is a multiinput mul-
controller in experiment, so a sample and hold (S/H) function is tioutput system. Using UC port as an example, if Io is defined
utilized between power stage and control in Fig. 7. as the converter output current at port II, then its outputs are îo
The current-fed UC-port can be regarded as a boost converter and V̂12 , inputs are φ̂12 and D̂1 . Let H11 represent the trans-
cascaded by a voltage-fed port. Therefore, in order to achieve fer function of Ifc controller, H22 is the transfer function of
fast response, IUC current control is selected as the inner loop to cascaded V12 controller and IUC controller; this two-input-two-
cascade with duty cycle D1 control loop to keep V12 constant. output system block diagram is shown in Fig. 8, where G12 and
G22 is the converter plant function that connected to V12 control G21 represent the interaction plant functions, G11 and G22 can
loop where G22A is the plant function between D1 and IUC , and be seen to Fig. 7. The small signal transfer functions in Fig. 8
G22B is the plant function between IUC and V12 . The advantages are shown below; it can be seen that the two outputs are coupled
of keeping V12 constant are as following: 1) ZVS conditions to each other
of THB can be satisfied under wide input voltage variations.


2) The transients (overshoot/undershoot) of inverter DC link îo G11 G12 φ̂12
= (17)
voltage become smaller resulting from fast system dynamics by V̂12 G21 G22 D̂1
the proposed duty cycle control cascaded with current control.
where
B. BU-port Control G11 = (0.5nCp L1 s2 − mgL1 s + nD12 )/(0.5Cp L1 s2 + D12 )
The BU-port control of Fig. 7 will implement the UC state- G12 = −g(−ILdc1 L1 s + D1 V12 )/(0.5Cp L1 s2 + D12 )
of-charge (SOC) control and keep V56 constant. The UC SOC
command is derived from the vehicle speed, which is not a G21 = −mL1 s/(0.5Cp L1 s2 + D12 )
constant value but a varied value according to the vehicle speed G22 = −(−IL dc1 L1 s + D1 V12 )/(0.5Cp L1 s2 + D12 )
to get ready for an oncoming acceleration or deceleration event
[16]. The control of UC SOC is realized by regulating φ32 . For n = (V12 (π−2φ12 )/4πωLr 12 ), m=(V34 (π−2φ12 )/4πωLr 12 ),
example, if UC SOC is lower than its reference, φ32 will increase g = φ12 (π − φ12 )/4πωLr 12 , Cp = 0.5C1 = 0.5C2 .
to provide more power from BU port, thus to adjust UC SOC It shows that G12 and G22 have one right-half-plane (RHP)
following its reference. However, φ32 is not only determined zero, which is 1 kHz at full load power and beyond this when
by the UC SOC control loop, it also needs to be synchronized load power decreases. This RHP zero makes theoretical decou-
with φ12 of UC port to avoid the circulation energy between pling network impossible since that will need exact cancella-
these two ports, this synchronization mechanism is highlighted tion of numerator and denominator. Pure static-decouplers (gain
898 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 27, NO. 2, FEBRUARY 2012

Fig. 9. (a) Bode plots of îo / φ̂ 1 2 (G 1 1 H 1 1 ) with and without controller, (b) bode plot of H 1 1 .

decouplers) may be a simple solution but it cannot solve RHP and phase margin. Besides the zero-pole pair of PI controller,
zero related stability problem either. the digital controller in the system also introduces another pole
In this paper, a new method is proposed to design the two at higher frequency. Although the controller H11 is mainly de-
individual controllers specifically to achieve actual decoupling signed according to G11 , it is shown in Fig. 9(b) that it has a
and stability. In Fig. 8, if the two controllers H11 and H22 are much reduced gain in selected frequency band to reduce cou-
integrated into the transfer functions, the whole system transfer pling factor.
functions become The V12 controller of UC port includes the inductor cur-



rent control as inner current control loop. The control-to-output
îo −1
G11 H11 G12 H22 w1
=A (18) transfer functions for both UC and BU ports inductor currents
V̂12 G21 H11 G22 H22 w2
are derived as below:
1 + G11 H11 G12 H22 îUC 2V12
where A = , ≈
G21 H11 1 + G22 H22 D̂1 D12 R12


îo 1 G11 H11 (1 + G22 H22 ) − G12 H22 G21 H11 (R12 C1 /4 + RC 12 C1 /2)s + 1
= ×
|A| (C1 L1 /2D12 )s2 + (RC 12 C1 /2 − (L1 /D12 R12 ))s + 1
V̂12 G21 H11
(21)
G12 H22  
w . îBU 2V56
G22 H22 (1 + G11 H11 ) − G12 H22 G21 H11 1 ≈
w2 D̂3 D32 R56
(19) (R56 C5 /4 + RC 56 C5 /2)s + 1
×
From this new transfer function it can be seen coupling factor (C5 L2 /2D32 )s2 + (RC 56 C5 /2 − (L2 /D32 R56 ))s + 1
becomes G12 H22 and G21 H11 , since H22 and H11 are integrator (22)
based controller, the design target is then changed to maintain-
ing a smaller gain (<1) for them above low-frequency band. where RC 12 and RC 56 are the ESR resistors for C12 and C56 ,
Furthermore, the bandwidth of G22 H22 and G11 H11 should respectively. C12 and C56 are the total DC capacitors at UC and
be designed according to system requirement, since G12 has BU side. R12 and R56 are equivalent load resistances at DC link
an RHP zero, the bandwidth of G12 H22 is below RHP zero sides for UC port and BU port.
frequency. The IBU and IUC controllers are designed based on (21) and
The small signal transfer function îo /φ̂12 is derived as (22). Their bandwidth is selected as 1/6 of switching frequency
based on [23], [24]. The voltage controller of UC port is designed
îo V12 (π − 2φ12 ) 1 based on the control-to-output function of V12 , which is derived
= (20)
φ̂12 4πωLr 12 n as
where n is transformer turns ratio between v2 and v1 , ω = 2πf, v̂12 D1 ((ωc1 L1 /D12 )||R12 ||R12 )

and f is switching frequency. v̂c Rf
Since Io = Ifc + Iload , the Ifc controller can be designed
(1 − (L1 /D12 R12 )s)((Rc12 C1 /2)s + 1)
based on (20). Fig. 9(a) shows the bode plots of îo /φ̂12 before × (23)
and after compensation. The controller parameters can be de- ((R12 C1 /4)s + 1)(s/ωc1 + 1)
signed accordingly by choosing appropriate bandwidth at 1 kHz where ω c 1 is the crossover frequency of IUC control loop.
WANG et al.: ASYMMETRICAL DUTY CYCLE CONTROL AND DECOUPLED POWER FLOW DESIGN 899

Fig. 10. (a) Bode plots of v̂ 1 2 /v̂ c (G 2 2 H 2 2 ) with and without controller, (b) bode plot of H 2 2 .

Bode plots of v̂12 /v̂c before and after compensation are shown
in Fig. 10(a). The crossover frequency (100 Hz) is set away from
the RHP zero frequency, which is beyond 1 kHz. As same as
the Io controller, the digital controller in the system also intro-
duces another pole at higher frequency. The voltage controller
of BU port can be derived similarly. Furthermore, it is shown
in Fig. 10(b) that H22 has a much reduced gain in selected
frequency band to reduce coupling factor.

IV. REAL TIME DIGITAL SIMULATION


AND EXPERIMENTAL VERIFICATIONS

To verify the proposed control method, the whole system


with the energy storage and this three-port DC-DC converter
for FC vehicle is designed and simulated using real time digital
simulator (RTDS) at urban ECE R40 driving cycle. The detail Fig. 11. Simulation results of vehicle speed, V U C , V B U , and BU SOC.
description of ECE R40 driving cycle can be found in [25].
The powerful parallel calculation capability of RTDS results in
fast and accurate enough simulation results. The simulation is
downscaled in order to compare with hardware experimental
results at similar power rating. The dynamic models of essential
sub-system components were presented in [22].
The simulation results of applying proposed asymmetrical
duty cycle were obtained from RTDS and compared with those
of using fixed duty cycle method. The simulation results are
shown in Figs. 11–14. Although the average models of the con-
verter are applied in the simulation, the converter conduction
loss was estimated approximately using a resistor whose loss
was modeled as the function of the phase shift angles, but switch-
ing loss cannot be modeled. It is observed that both methods can
achieve similar system performances for VUC , BU SOC, IBU ,
and IUC . The differences between two methods are among FC
average power, V12 , V56 , as well as inverter DC link voltage. Fig. 12. Simulation results of load power profile, FC power, and inverter DC
Fig. 11 shows that both methods have similar VUC , and BU link voltage.
SOC; Fig. 12 shows FC average power of proposed method
is 8% lower than that using fixed duty cycle, it implicit lower of two methods can be observed as 77.6 kJ for proposed method
converter loss in proposed method. Since constant V12 and V56 and 84.6 kJ for fixed duty cycle method under the given driving
resulting in smaller phase shift angles and lower transformer cycle. Therefore, the proposed method can achieve the opti-
currents, thus lower rms loss. In the simulation, the FC energy mized maximum-fuel economy design for FC vehicle. Fig. 13
900 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 27, NO. 2, FEBRUARY 2012

Fig. 13. Simulation results of UC-port: D 1 , V1 2 , and IU C .


Fig. 15. Simulation results: P 1 2 + P 2 3 + P 3 1 in ECE40 driving cycle.

Fig. 14. Simulation results of BU-port: D 3 , V5 6 , and IB U .


Fig. 16. Test bed of emulated fuel cell vehicle power train in the laboratory.

shows both methods have similar IUC , however, V12 is con-


PI control parameters of Ifc controller is designed as: Kp = 0.2,
stant in proposed ADC but varied in FDC. Fig. 14 shows both
Ki = 1000. The control parameters of two voltage controllers
methods have similar IBU , however, V56 is constant in proposed
for V12 and V56 are selected as: Kp = 0.04, Ki = 3000; while Kp
ADC but varied in FDC.
= 0.01, Ki = 10000 are the control parameters for two current
To further describe the relation between minimum circulating
controllers. The PI parameters of UC SOC controller are: Kp =
loss and phase shift operating principle, P12 + P23 + P31 and
0.04, Ki = 1 when 9 V < V uc < 13 V; Kp = 0.025, Ki = 1 when
their respective values are derived in Fig. 15. P12 + P23 + P31
6 V < V uc < 9 V. In this paper, vehicle inertia of motor drive
is always equal to zero except in the range where UC voltage
of a mid-size car is derived based on equation (24) from [26]:
falls too low thus not matching to another two ports.
The experimental setup to verify the design is shown in  2
r
Fig. 16. Fuel cell has been emulated by a 40 V battery pack J =M (24)
ig
and a 1:8 boost DC-DC converter (to boost voltage to 320 V).
The battery is BP20-12 (12 V, 20 Ah) from B.B. Battery Corp. where M is mass, r is the wheel radius and ig is the gear
and UC is BMOD0430E16 (16 V, 430 F) from Maxwell. In the ration. J is calculated as 1.2 kgm2 when M = 1200 kg, r =
experiments, the ultracapacitor voltage varies from 7 to 13 V. 0.3 m, ig = 9. The flywheel inertia in the experimental system
A 2.5 kW three-port THB bidirectional DC-DC converter was is downscaled as 0.057 kgm2 to emulate a smaller vehicle. It
built in the laboratory to interface the low voltage battery and is made by SM45 C steel, the radius is 100 mm, the height is
UC. Switching frequency of the converter is 20 kHz. This THB 46 mm. Since the load torque generated by flywheel is already
converter is connected to the inverter to drive a 10 kW PMSM close to desired torque-speed profile, the generator in system is
dynamometer with flywheel. The high-voltage bus of the in- open loaded.
verter is controlled to be 320 V in the experiment. The flywheel Figs. 17 and 18 showed the experimental results of THB con-
is utilized to reflect vehicle mass/inertia, The control design verter steady state voltage and current waveforms at boost mode
was implemented using a digital-signal-processor (DSP). The and buck mode under different duty cycle control methods. The
WANG et al.: ASYMMETRICAL DUTY CYCLE CONTROL AND DECOUPLED POWER FLOW DESIGN 901

Fig. 17. Experimental results of THB converter voltage and current waveforms at boost mode under different duty cycle methods: (a) FDC, (b) SDC, and
(c) ADC.

Fig. 18. Experimental results of THB converter voltage and current waveforms at buck mode under different duty cycle methods: (a) FDC, (b) SDC, and
(c) ADC.

Fig. 19. THB converter efficiency measurement at boost mode under different Fig. 20. THB converter efficiency measurement at buck mode under different
control methods when V in1 = 10.7 V, V in3 = 11.45 V, V 3 4 = 310 V. control methods when V in1 = 9.5 V, V in3 = 12.8 V, V 3 4 = 300 V.
902 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 27, NO. 2, FEBRUARY 2012

Fig. 21. Experimental results of fixed duty cycle: (a) Load power profile, inverter DC bus voltage, and Ifc ; (b) V 1 2 , V 5 6 , D 1 , and D 3 ; (c) V U C , V B U , IU C , and
IB U .

Fig. 22. Experimental results of proposed varied duty cycle method: (a) Load power profile, inverter DC bus voltage, and Ifc ; (b) V 1 2 , D 1 , V 5 6 , and D 3 ;
(c) V U C , V B U , IU C , and IB U .

input conditions are V in1 = 10.7 V, V in3 = 11.45 V, V34 = Figs. 21 and 22 showed the experimental results using proposed
310 V for boost mode, and V in1 = 9.5 V, V in3 = 12.8 V, V34 = method and fixed duty cycle method under the same load power
300 V for buck mode. The three port voltages are chosen based profile. The load power profile is generated by PMSM motor
on one typical case in vehicle drive application: UC voltage is drive emulating vehicle drive and shown in Fig. 21(a). Ifc of
lower than BU and referred inverter DC bus voltage to the pri- fixed duty cycle is 9 A, while it is 8 A in the proposed method,
mary side of transformer. Three different control methods are FC average power of proposed method is 12.5% lower than that
compared. From Fig. 17 it can be seen that both FDC and SDC using fixed duty cycle, since UC and BU SOC are controlled
method cannot maintain ZVS condition for S2 at boost mode, back to original value, as a result the total fuel economy has been
since their transformer current has a down slope thus little or improved 12.5% by using proposed control for THB, which is
no margin for a positive switch current. However, ADC method consistent with the simulated results. The V12 is regulated as
still can maintain ZVS since its transformer current has an up constant 26 V if proposed duty cycle control is applied other-
slope in Fig. 17(c). Fig. 18 shows three methods at buck mode, wise it will follow twice the voltage variations of ultracapacitor.
although FDC can maintain ZVS for S1, it has less ZVS margin Although VUC , IUC , and inverter DC link voltage of the two
of S1 compared to ADC methods. While for SDC method, it methods are almost same, the converter control variables in-
cannot maintain ZVS for S3/S4 at buck mode. The efficiency cluding phase shift angles and duty cycle are totally different
measurements of different methods at boost and buck mode are so converter performance is optimized and the converter loss is
shown in Figs. 19 and 20 where a typical application is selected minimized by using proposed method.
such that the power distribution between three port in Fig. 19 To verify the effect of proposed phase shift and duty cycle
are controlled as P1 = P3 , P1 + P3 = Pin , P2 = −P o; and P1 controller on this coupled system, the system is tested and com-
= P3 , P1 + P3 = −P o, P2 = Pin in Fig. 20. It can be observed pared with proposed controller with different bandwidth and
that the ADC method has highest efficiency in either boost or another controller. The result is shown in Fig. 23. It can be seen
buck mode. that the transient of Io will not affect V12 with proposed con-
To verify the proposed method in dynamic load condition, troller design; however, V12 has a big undershoot when using
the whole system is tested under urban driving cycle ECE 40. another controller design.
WANG et al.: ASYMMETRICAL DUTY CYCLE CONTROL AND DECOUPLED POWER FLOW DESIGN 903

Fig. 23. Experimental results: (a) Both H 1 1 and H 2 2 bandwidth at 100 Hz. (b) Proposed one: H 1 1 bandwidth 1 kHz, H 2 2 bandwidth 100 Hz.

V. CONCLUSION plications,” IEEE Trans. Ind. Appl., vol. 27, no. 1, pp. 63–73, Jan./Feb.
1991.
To interface the hybrid energy storage of low voltage battery [7] M. H. Kheraluwala, R. W. Gascoigne, D. M. Divan, and E. D. Baumann,
and ultracapacitor to the high-voltage bus in a fuel cell vehi- “Performance characterization of a high-power dual active bridge DC-to-
DC converter,” IEEE Trans. Ind. Appl., vol. 28, no. 6, pp. 1294–1301,
cle, a C-C-V three-port triple-half-bridge bidirectional DC-DC Nov./Dec. 1992.
converter with a new asymmetrical duty cycle control method [8] K. Vangen, T. Melaa, and A. K. Adnanes, “Soft-switched high-frequency,
is presented in this paper. The proposed asymmetrical duty cy- high power DC/AC converter with IGBT,” in Proc. IEEE Power Electron.
Spec. Conf., Toledo, Spain, Jun./Jul.1992, pp. 26–33.
cle control not only can achieve a wide ZVS range for pre- [9] F. Z. Peng, H. Li, G. J. Su, and J. Lawler, “A new ZVS bidirectional DC-
sented converter when ultracapacitor voltage has a wide varia- DC converter for fuel cell and battery application,” IEEE Trans. Power
tion range, but also reduce converter peak current and rms cur- Electron., vol. 19, no. 1, pp. 54–65, Jan. 2004.
[10] J. L. Duarte, M. Hendrix, and M. G. Simoes, “Three-port bidirectional
rent resulting in lower rms loss. In addition, the laws regarding converter for hybrid fuel cell systems,” IEEE Trans. Power Electron.,
minimum circulation losses between multiports are also studied vol. 22, no. 2, pp. 480–487, Mar. 2007.
and it is naturally achieved by phase shift operating principle. [11] C. Zhao, S. D. Round, and J. W. Kolar, “An isolated three-port bidirectional
DC-DC converter with decoupled power flow management,” IEEE Trans.
Furthermore, the dynamic model of this converter with multiple Power Electron., vol. 23, no. 5, pp. 2443–2453, Sep. 2008.
control variables was developed so the proposed phase shift and [12] H. Tao, A. Kotsopoulos, J. L. Duarte, and M. A. M. Hendrix, “Transformer-
duty cycle control can be designed with reduced coupling factor. coupled multiport ZVS bidirectional DC-DC converter with wide input
range,” IEEE Trans. Power Electron., vol. 23, no. 2, pp. 771–781, Mar.
Therefore, the proposed technology can achieve optimum per- 2008.
formance for low and variable voltage, high current application [13] H. Tao, J. L. Duarte, and M. A. M. Hendrix, “Three-port triple-half-bridge
in this paper. Finally, both RTDS simulation and experimental bidirectional converter with zero-voltage switching,” IEEE Trans. Power
Electron., vol. 23, no. 2, pp. 782–792, Mar. 2008.
results validated that proposed asymmetrical duty cycle method [14] D. Liu and H. Li, “A ZVS bidirectional DC–DC converter for multiple
could achieve higher efficiency than other duty cycle methods in energy storage elements,” IEEE Trans. Power Electron., vol. 21, no. 5,
both steady state and varied driving cycle. The reduced coupling pp. 1513–1517, Sep. 2006.
[15] D. Xu, C. Zhao, and H. Fan, “A PWM plus phase-shift control bidirectional
factor between phase shift control and duty cycle control was DC-DC converter,” IEEE Trans. Power Electron., vol. 19, no. 3, pp. 666–
also validated in the experiment based on the emulated fuel cell 675, May 2004.
power train test bed built in the laboratory. [16] Moreno, M. E. Ortuzar, and J. W. Dixon, “Energy-management system
for a hybrid electric vehicle, using ultracapacitors and neural networks,”
IEEE Trans. Ind. Electron., vol. 53, no. 2, pp. 614–623, Apr. 2006.
REFERENCES [17] M. N. Gitaua, G. Ebersohna, and J. G. Kettleboroughb, “Power processor
for interfacing battery storage system to 725 V DC bus,” Energy Convers.
[1] A. F. Burke, “Batteries and ultracapacitors for electric, hybrid, and fuel Manag., vol. 48, no. 3, pp. 871–881, Mar. 2007.
cell vehicles,” Proc. IEEE, vol. 95, no. 4, pp. 806–820, Apr. 2007. [18] Z. Wang and H. Li, “Optimized operating mode of current-fed dual half
[2] C. C. Chan, “The state of the art of electric, hybrid, and fuel cell vehicles,” bridges DC-DC converters for energy storage applications,” in Proc.
Proc. IEEE, vol. 95, no. 4, pp. 704–718, Apr. 2007. IEEE Energy Convers. Congr. Expo., San Jose, CA, Sep. 2009, pp. 731–
[3] J. M. Miller, U. Deshpande, T. J. Dougherty, and T. Bohn, “Power elec- 737.
tronic enabled active hybrid energy storage system and its economic vi- [19] L. Solero, A. Lidozzi, and J. A. Pomilio, “Design of multiple-input power
ability,” in Proc. IEEE Applied Power Electron. Conf., Washington, DC, converter for hybrid vehicles,” IEEE Trans. Power Electron., vol. 20,
Feb. 2009, pp. 190–198. no. 5, pp. 1007–1016, Sep. 2005.
[4] K. Wang, C. Y. Lin, L. Zhu, D. Qu, F. C. Lee, and J. S. Lai;, “Bi-directional [20] P. Thounthong, S. Rael, B. Davat, and I. Sadli, “A control strategy of
DC to DC converters for fuel cell systems,” Power Electron. Transp., fuel cell/battery hybrid power source for electric vehicle applications,” in
pp. 47–51, Oct. 1998. Proc. IEEE Power Electron. Spec. Conf., Jeju, South Korea, Jun. 2006,
[5] G.-J. Su and F. Z. Peng, “A low cost, triple-voltage bus DC-DC converter pp. 1–7.
for automotive applications,” in Proc. IEEE Appl. Power Electron. Conf., [21] H. Li and D. Liu, “Power distribution strategy of fuel cell vehicle system
Austin, TX, Mar. 2005, pp. 1015–1021. with hybrid energy storage elements using triple half bridge (THB) bidi-
[6] R. W. DeDonker, D. M. Divan, and M. H. Kheraluwala, “A three-phase rectional DC-DC converter,” in Proc. IEEE Ind. Appl. Soc. Annu. Meeting,
soft- switched high power density DC-DC converter for high power ap- New Orleans, LA, Sep. 2007, pp. 636–642.
904 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 27, NO. 2, FEBRUARY 2012

[22] L. Wang and H. Li, “Maximum fuel economy-oriented power management Zhan Wang (S’09) received the B.S. and M.S. de-
design for a fuel cell vehicle using battery and ultracapacitor,” IEEE Trans. grees in electrical engineering from Huazhong Uni-
Ind. Appl., vol. 46, no. 3, pp. 1011–1020, May/Jun. 2010. versity of Science and Technology, Wuhan, China, in
[23] R. D. Middlebrook, “Modeling current-programmed buck and boost reg- 2002 and 2005, respectively. He is currently working
ulators,” IEEE Trans. Power Electron., vol. 4, no. 1, pp. 36–52, Jan. toward the Ph.D. degree in electrical engineering at
1989. Florida State University, Tallahassee, FL.
[24] L. H. Dixon, “Average current mode control of switching power supplies,” From 2005 to 2007, he worked with SANTAK
Unitrode Power Supply Design Seminar, p. 7–17, 1991. Electronics (Shenzhen) Co., Ltd. in Shenzhen, China.
[25] P. Corbo, F. E. Corcione, F. Migliardini, and O. Veneri, “Experimental His current research interest is multiport, multiphase
assessment of energy-management strategies in fuel cell propulsion sys- bidirectional DC-DC converter, and power electron-
tems,” J. Power Sources, pp. 799–808, Jul. 2006. ics applications in renewable energy generation.
[26] B. Chen, Y. Gao, M. Ehsani, and J. M. Miller, “Design and control of
a ultracapacitor boosted hybrid fuel cell vehicle,” in Proc. IEEE Vehicle
Power Propulsion Conf., Dearborn, MI, Sep. 2009, pp. 696–703.

Hui Li (S’97–M’00–SM’01) received the B.S. and


M.S. degrees from Huazhong University of Science
Lei Wang (S’09–M’11) received the B.S. degree and Technology, China in 1992 and 1995, respec-
from Shanghai Jiaotong University, Shanghai, China, tively, and the Ph.D. degree from the University
in 1996, and M.S. degree from Zhejiang University, of Tennessee, Knoxville, in 2000, all in electrical
Hangzhou, China, in 2002, both in electrical engi- engineering.
neering. He received the Ph.D. degree in electrical She is currently an Associate Professor in the
engineering from Florida State University, in 2010. Electrical and Computer Engineering Department at
He is currently with GE Global Research at the Florida A&M University – Florida State Uni-
Niskayuna, NY. His research interests include mod- versity College of Engineering, Tallahassee, FL. Her
eling and control of bidirectional DC-DC converters, research interests include bidirectional DC-DC con-
fuel cell hybrid vehicle, and plug-in hybrid electric verters, cascaded multilevel inverters, and power electronics application in hy-
vehicle. brid electric vehicles.

You might also like