Aalborg Universitet
Open-Circuit Fault Analysis and Fault-Tolerant Control for 2/3-Level DAB Converters
Song, Chaochao; Yang, Yongheng; Sangwongwanich, Ariya; Blaabjerg, Frede
Published in:2021 IEEE 12th Energy Conversion Congress & Exposition - Asia (ECCE-Asia)
DOI (link to publication from Publisher):10.1109/ECCE-Asia49820.2021.9479285
Publication date:2021
Document VersionEarly version, also known as pre-print
Link to publication from Aalborg University
Citation for published version (APA):Song, C., Yang, Y., Sangwongwanich, A., & Blaabjerg, F. (2021). Open-Circuit Fault Analysis and Fault-TolerantControl for 2/3-Level DAB Converters. In 2021 IEEE 12th Energy Conversion Congress & Exposition - Asia(ECCE-Asia) (pp. 696-701). [9479285] IEEE Press. https://doi.org/10.1109/ECCE-Asia49820.2021.9479285
General rightsCopyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright ownersand it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.
- Users may download and print one copy of any publication from the public portal for the purpose of private study or research. - You may not further distribute the material or use it for any profit-making activity or commercial gain - You may freely distribute the URL identifying the publication in the public portal -
Take down policyIf you believe that this document breaches copyright please contact us at [email protected] providing details, and we will remove access tothe work immediately and investigate your claim.
Downloaded from vbn.aau.dk on: February 05, 2022
Open-Circuit Fault Analysis and Fault-Tolerant Control for 2/3-Level DAB Converters Chaochao Song*, Yongheng Yang+, Ariya Sangwongwanich*, and Frede Blaabjerg*
*Department of Energy Technology, Aalborg University, Aalborg, Denmark +College of Electrical Engineering, Zhejiang University, Hangzhou, China
E-mails: [email protected], [email protected], [email protected], [email protected]
Abstract—Open-circuit faults (OCFs) on power switches are crucial issues for the two-three (2/3)-level dual-active-bridge (DAB) DC-DC converters, resulting in various performance degradation such as DC bias, overcurrent and capacitor voltage imbalance. There are two necessary steps for overcoming these problems, i.e., fault diagnosis and fault-tolerant control. In order to identify the faulty switch, the characteristics of the transient waveforms when the OCF occurs on each switch are analyzed in this paper. Based on the analysis, the midpoint voltage of each bridge arm is employed as the diagnosis signals. According to the mean values and duty cycles of the midpoint voltages, the faulty switch can be located accurately. Subsequently, a fault-tolerant control method called “complementary-switch-blocking” (CSB) is proposed through modulation reconfiguration. In the proposed CSB method, when one switch breaks down, the gate-driving signal of its complementary switch is blocked and the OCF effects can be offset. Finally, simulation results demonstrate that the OCF effects can be reduced significantly with the proposed fault-tolerant method, and the power transmission capacity can be improved compared with the traditional fault-tolerant method.
Keywords—DAB converter, open-circuit fault, fault diagnosis, fault-tolerant control.
I. INTRODUCTION Solar energy is one of the fastest growing renewable
energy driven by the strong demand for renewable energy, reduction in cost of photovoltaic (PV) modules, and the development of the PV technologies [1], [2]. Recently, medium-voltage DC (MVDC) structure is considered a promising solution for large-scale PV plants to increase the power capacity and improve the performances with higher efficiency and flexibility, and lower control complexity [3]. In the MVDC PV system, the interfacing DC-DC converter between the low-voltage DC (LVDC) bus and MVDC bus plays a key role, which should withstand high rated voltage and achieve high step-up ratio. Dual-active-bridge (DAB) DC-DC converters have a wide range of applications in the LVDC systems, due to the advantages such as high power density, current isolation and soft-switching capability. However, limited by the performance and reliability of MV power semiconductors, it is challenging to interface the MVDC network with a traditional two-level DAB converter. Compared with two-level DAB converters, the two-three (2/3)-level DAB converters, as shown in Fig. 1, have a higher voltage blocking capability, and provide more control degrees of freedom (DoFs) to further improve the performance with the neutral point-clamped (NPC) bridge in the MV side. Therefore, 2/3-level DAB converters have a high potential to be applied in medium-voltage DC systems [4].
Due to an increased number of power devices for the MV power electronics systems, ensuring their reliability is very crucial. According to previous studies, one of the most vulnerable component in power electronics applications is the power semiconductors [5]. Beside, the gate drivers are also a factor which is prone to failure. As a result, open-circuit faults
(OCFs) for the converters account for a large proportion of reliability research. When the OCFs occurs on the DAB converters, the terminal voltage waveforms of the transformer will be asmmetrical, resulting in the DC bias, overcurrent, capacitor voltage imbalance and other issues. These effects will increase the voltage and current stresses on the power devices, and threaten the safety and reliability of the DAB converters with long term operation [6]. In order to overcome these issues, two steps should be taken, i.e., fault diagnosis and fault-tolerant control.
For fault diagnosis, the methods require simple, fast, and accurate detection. Some fault diagnosis methods have been proposed based on the residual analysis, circuit analysis, and DC components of phase currents for the two-level DAB converters, matrix converters, and three-phase DAB converters, respectively [7]–[9]. However, they can merely determine the faulty bridge arm rather than the accurate switch. In [10]–[12], fault diagnosis methods are applied based on the midpoint voltage of each bridge arm. However, these methods are proposed for the two-level DAB converters. When the midpoint voltages are employed as the diagnosis signals for the 2/3-level DAB converters, the diagnosis process should be redesigned, because the midpoint voltage waveform in the NPC bridge is different from that in the two-level full bridge.
After fault diagnosis has been carried out, the fault-tolerant control method should be applied to eliminate the OCF effects and ensure safe operation of the DAB converters. In [9], a fault-tolerant control method was proposed for the three-phase DAB converters. When the faulty bridge arm is located, it will be disabled and the converters will operate in single-phase state. This method is simple but not suitable for the single-phase DAB converters due to its limited redundancy. A fault-tolerant method called “primary side lower power - secondary side bypass arm” (PLP-SBA) has been proposed and applied to the two-level DAB converters [11], [12]. In this method, when the OCF occurs on the switch in the primary side, the transferred power is reduced to ensure that the maximum inductor current is within the rated value. On the other hand, if the OCF occurs on the switch in the
Fig. 1. A two-three-level dual-active-bridge converter.
1 : n
Ls
S13S11
S14S12
S22
S23
iL
i1
vab vcdC1
D11 D13
D12 D14
D22
D23
S21
S24
S26
S27
D26
D27
S25
S28
C2
C3
V2
i2
D21
D24
D25
D28
D1
D2
D3
D4
a
b
c
d
+LV side
MV side
oio
g
secondary side, the gate-driving signal of the other switch which is in the same arm with the faulty one is blocked to maintain the symmetry of the waveforms and eliminate the performance degradation caused by the OCFs. This method can be applied to the 2/3-level DAB converters. However, due to larger number of switches, the 2/3-level DAB converters have more potential to enhance the power range and improve the power capacity during fault-tolerant operation by modulation reconfiguration.
Since the OCF diagnosis and fault-tolerant control have not been addressed for the NPC-based DAB converters, this paper proposes a fault diagnosis method and a fault-tolerant control method for the 2/3-level DAB converters. In the proposed fault diagnosis method, the midpoint voltages are chosen as the diagnosis signals. By calculating the mean values and duty cycles of the midpoint voltages, the faulty switch can be located accurately. Furthermore, the fault-tolerant control strategy called “complementary-switch-blocking” (CSB) method is proposed. In the proposed method, when the OCF occurs on a switch, the gate-driving signal of its complementary switch is blocked, and the OCF effects can be eliminated, ensuring the fault-tolerant operation of the DAB converters by modulation reconfiguration. The rest of this paper is organized as follows. In Section II, the faulty modes and fault diagnosis method is introduced. In Section III, the CSB method is proposed, along with the comparison with the traditional fault-tolerant method. The simulation results are given in Section IV, and a conclusion is provided in Section V.
II. OPEN-CIRCUIT FAULT ANALYSIS AND FAULT DIAGNOSIS
A. OCF effects and fault modes analysis As shown in Fig. 1, V1 and V2 are the two DC-link voltages.
vab and vcd are the AC terminal voltages of the isolated transformer with the turns ratio 1: n. Ls is the series inductor, and iL denotes the inductor current of Ls. o is the midpoint of the two capacitors C2 and C3, and g is the negative output point
of the NPC bridge. Phase-shift control is the most popular control scheme for the DAB converters, as shown in Fig. 2, in which Ths is a half switching cycle. The magnitude of the transferred power is adjusted by the phase-shift ratio D between the two bridges, whose range is 0 < D ≤ 0.5. From Fig. 2, it can be seen that in the normal state, the waveforms of the voltages and inductor current are symmetric over one switching cycle. Therefore, there is no DC bias in iL. However, when the OCF occurs on the switch in the NPC leg, the symmetry will no longer be maintained, and the waveforms of vcd and iL will be affected. Note that the OCF analysis for the full bridge in the LV side can be achieved in the same way as it has been done for the two-level DAB converters. Therefore, we focus on the OCFs in the MV side in this paper.
To analyze the distortion of the waveforms caused by OCFs clearly, the detailed switching characteristics when the OCF occurs on S21 are discussed, as shown in Fig. 3. In Fig. 3, vcg and vdg are the midpoint voltages of the two bridge arms in the MV side, and t0 is the first zero-crossing point of iL during one switching cycle. It can be seen that compared with the normal state, the waveform of vcd is distorted during the interval [A, B], whose value reduces from V2 to 0.5V2. This is because during the interval [A, B], iL becomes negative. During the normal operation, where S21 can function properly, the current will flow through S21, S22, S27 and S28, as shown in Fig. 4 (a). However, when the OCF occurs on S21, the current draws from the midpoint o, will flow through D1 instead of S21, as shown in Fig. 4 (b). In this condition, the value of vcd decreases to 0.5V2. Furthermore, in the normal state, the waveforms of the midpoint voltages vcg and vdg are square waves with 50% duty cycle. However, during the faulty interval [A, B], from Fig. 4 (b), it can be seen that the value of vcg also decreases from V2 to 0.5V2.
According to the fault anlysis, the waveform of vcd is not symmetric in one switching cycle. The slope and value of the inductor current iL is determined by the two voltages vab and vcd, which can be seen from the expression of iL
S22
S23
S24
S26
S27
S25
S28
C2
C3
V2
i2
D1
D2
D3
D4
c
d
+
o
g
S21
(a) (b) Fig. 4. Current conduction paths for the MV side at the interval [A, B] when: (a) S21 works normally. (b) the OCF occurs on S21.
v ab
v cd
(100
V/d
iv)
i L(2
0A/d
iv)vab
vcd iL
Normal state OCF state 0.1 ms/div
Fig. 5. Waveforms when the OCF occurs on S21 with the simulation parameters: input voltage V1 = 200 V, reference output voltage V2ref = 300 V, inductor Ls = 100 μH, and switching frequency fs = 10 kHz.
S22
S23
S24
S26
S27
S25
S28
C2
C3
V2
i2
D1
D2
D3
D4
c
d
+
o
g
S21Ths
vcd
iL tDThs
S11 S14
S21 S22 S27 S28 S23 S24 S25 S26
S12 S13
vab
Fig. 2. Phase-shift control scheme for the 2/3-level DAB converters.
t
vcg
vdg
Normal state OCF on S21
A B
vcd
iL
V20.5V2
V20.5V2
DThs
vab t0
A= (1+t0)Ths, B= (1+D)Ths0
Fig. 3. Typical waveforms when the OCF occurs on S21.
( ) [ ( ) ( ) / ]L ab cd sdi t dt v t v t n L (1) Thus, the waveform of iL will also become asymmetric in faulty state. This result is that the DC bias in iL, as shown in Fig. 5, causing higher peak current and potential saturation for the transformer. Besides, it can be seen from Fig. 4 (b) that during the faulty interval, the midpoint current io draws from the midpoint o. However, there is no interval where the current injects into the midpoint o in one switching cycle, which leads to capacitor voltage imbalance, and increase the voltage stress for some power devices.
Similarly, when the OCF occurs on other switches in the MV side, the faulty-state waveforms can be obtained. Note that when the value of the voltage conversion ratio k (i.e., k = nV1/V2) is in different ranges: 0 < k ≤ 0.5, 0.5 < k ≤ 1, and k ≥1, the waveform of vcd will be different under the OCF state. In this paper, only 0.5 < k ≤ 1 is considered, but the same analysis can also be applied to the other conditions. Fig. 6 shows the faulty-state waveforms when the OCF occurs on one of the eight possible switches in the MV side when 0.5 < k ≤ 1. From Fig. 6, it can be seen that the waveform of vcd will be distorted for all cases. However, the waveforms of vcd when the OCFs occur on S21 and S28, S22 and S27, S23 and S26, S24 and S25 are identical, respectively. Therefore, if the waveform of vcd is used as the diagnosis signal to identify the faulty switch, it cannot differentiate which of the two switches has OCF. Fortunately, if the midpoint voltages vcg and vdg are used as the diagnosis signals, the faulty switch can be located accurately, because the midpoint voltages are different from each other for the eight switches. In this paper, the OCF diagnosis method is proposed based on the midpoint voltages vcg and vdg.
B. Fault Diagnosis method The structure of the proposed diagnosis method is shown
in Fig. 7. There are three steps to locate the faulty switch as follows.
Step 1: From Fig. 6, it can be seen that if the faulty switch is on the first bridge arm, the waveform of vcg will be distorted. Otherwise, if the faulty switch is on the second bridge arm, the waveform of vdg will be distorted. Based on this, the faulty switch can be located at a certain arm. Because the faulty conditions when the OCF occurs on the first arm and the
second arm are similar, only the condition for the fault in the first arm is analyzed in detail in the following two steps.
Step 2: When the mean value of vcg is less than 0.5V2 - α, the faulty switch can be located on S21 or S22. Otherwise, when it is larger than 0.5V2 + α, the faulty switch can be located on S23 or S24. Note that α is the threshold voltage to regulate the sensitivity of the diagnosis method and avoid false operation.
Step 3: The faulty switch is located accurately by waveforms transition. The main aim of the waveforms transition is to convert the waveforms of vcg to square waveforms and differentiate them according to their duty cycles. The details of the waveforms transition can be seen in Fig. 7 (b). After the waveforms transition, the faulty switch can be differentiated between S21 and S22, as well as between S23 and S24 based on the duty cycle of the final signal v0c. It should be noted that β is also a threshold value for calculating the duty cycle of the final signal v0c, which is mainly determined by the dead time.
III. PROPOSED FAULT-TOLERANT CONTROL METHOD After identifying the faulty switch, the fault-tolerant
control scheme should be employed to overcome the OCF issues. In previous research about the secondary-side-bypass-arm (SBA) method, the transferred power range was only considered when k = 1. Therefore, in this section, both the SBA method and the proposed CSB method will be analyzed, to make a comprehensive comparison.
A. Traditional SBA method In the SBA method, when one of the four switches in a
bridge arm is detected to be faulty, the gate-driving signals of the other three switches should be blocked to overcome the OCF effects. Fig. 8 gives the equivalent structures of the NPC bridge after employing the SBA method. For such equivalent structures, the steady-state waveforms of vab, vcd and iL during one switching cycle are shown in Fig. 9. Fig. 9 (a) and (b) are the waveforms in high power range and low power range, respectively, which are defined as Mode 1 and Mode 2. Due to the symmetry of iL in one switching cycle, and according to the expression of iL as shown in (1), the expressions of t0 of Mode 1 and Mode 2 can be obtained as
0
0
( 1 ) (2 1), Mode 1(1 ), Mode 2
hs
hs
t k D T kt DT k
(2)
The unified expression of the power P can be described as (3), and the normalized power P0 can be calculated as (4), where PN = V1V2Ths/4nLr is the maximum power.
0
1 ( ) ( )hsT
ab Lhs
P v t i t dtT
(3)
2 20 0
0 2 20 0
2( 2 2 1), Mode 1
2[( 1) 2 ], Mode 2N
t kt D D kPPP k t Dt D
(4)
From Fig. 9, it can be seen that the range of t0 in Mode 1 is 0 ≤ t0 ≤ D, and in Mode 2, D ≤ t0 ≤ 1. Combining the ranges with (2) and (4), the power range by using the SBA method can be obtained as
2 2 20
20
2 2 (4 4 1) (2 1) , Mode 1
0 2 2 , Mode 2
k k P k k k
P k k (5)
Fig. 6. Faulty-state waveforms when the OCF occurs on the eight switches in the MV side.
t
vcg
vdg
OCF on S21 t
vcg
vdg
OCF on S22 t
vcg
vdg
OCF on S23
t
vcg
vdg
OCF on S24 t
vcg
vdg
t
vcg
vdg
t
vcg
vdg
OCF on S25 OCF on S26
OCF on S27 t
vcg
vdg
OCF on S28
vab
vcd
iL
m = 0.5V2
m
m
m
m
B. Proposed CSB method Due to more power switches in the NPC bridge, there are
more possible current conduction paths for the fault-tolerant control schemes with potential to improve performances. According to the above analysis, the root cause of the OCF effects is that the faulty intervals introduce the asymmetry of the waveform of vcd. Therefore, in order to avoid these conditions, the waveform of vcd should become symmetric by using the fault-tolerant control scheme. For instance, from Fig. 3, it can be seen that when the OCF occurs on S21, the waveform of vcd will be distorted at the interval [(1+t0)Ths, (1+D)Ths], where the value of vcd decreases from +V2 to +0.5V2, along with the positive DC bias in iL. If the value of vcd can be increased from -V2 to -0.5V2 during the interval [t0Ths, DThs], the impact on the waveforms distortion of vcd and iL will counteract with the OCF on S21. Thus, the waveforms of vcd and iL will remain symmetric, and the DC bias in the current iL will be eliminated, as shown in Fig. 10.
From Fig. 6, it can be seen that when the OCF occurs on S24 or S25, the impact is similar with the above description. Therefore, when S21 is identified as the faulty switch, blocking the gate-driving signal of S24 or S25 can reduce the impact on the waveforms distortion of vcd and iL. When the OCF occurs on S21, from Fig. 4 (b), it can be seen that at the faulty interval [(1+t0)Ths, (1+D)Ths], the current i0 draws from the midpoint o. Fig. 11 (a) and (b) show the current flow paths during the interval [t0Ths, DThs] when the gate-driving signal of S24 and S25 is blocked, respectively. From Fig. 11 (a), it can be seen that the current i0 will inject into midpoint o. Therefore, when choosing S24 as the complementary switch to S21, the total charge injected into and drawn from the midpoint o will be equal to zero in one switching cycle, and the capacitor voltage balancing state can be guaranteed. However, from Fig. 11 (b), it can be seen that if S25 is chosen as the complementary switch to S21, the current i0 will also draw from o, which will accelerate the capacitor voltage imbalance. Therefore, S21 and S24 are a pair of complementary switches for fault-tolerant control. When one of S21 and S24 is detected to be faulty, the gate-driving signal of another one should be blocked. Similarly, S22 and S23, S25 and S28, S26 and S27 are other three complementary-switch pairs. Noted that when the gate signals of S22 and S23, or S26 and S27 are blocked, there is no current flow path on the four switches in the bridge arm. Therefore, the switching characteristics are similar to those in the SBA method. Thus, for the proposed CSB method, only two complementary pairs S21 and S24, S25 and S28 need to be further discussed.
When the OCF occurs on S21 or S24, and S25 or S28, the equivalent structures of the NPC bridge after employing the proposed CSB method are shown in Fig. 12. The steady-state waveforms during one switching cycle are shown in Fig. 13. Fig. 13 (a) and (b) are the waveforms in high power range and
(a) (b) Fig. 7. OCF diagnosis method based on the midpoint voltages. (a) Structure of the proposed diagnosis method. (b) Waveforms transition for the first bridge arm.
mean 20.5 ?cgv VY
213cg cgv v V mean0
1 v0c0 0.5 ?cv N
S21
Y S22N
vcg v0c
20.5 ?cgv VY
223cg cgv v V mean0
1 v0c0 0.5 ?cv
YS23
N S24
v0c
vdgmean 20.5 ?dgv V
Y2
13dg dgv v V mean0
1 v0d0 0.5 ?dv N
S25
Y S26N
vdg v0d
20.5 ?dgv VY
223dg dgv v V mean0
1 v0d0 0.5 ?dv Y
S27
N S28
v0d
N
*
*
*
*
vcg
V2
V2/2V2/3
d = 50%0
2V2/3V2/6 0
-V2/3
1
0
V2
V2/3
d < 50%0
2V2/3
0-V2/3
1
0
V22V2/3
d > 50%0
V2/30
-2V2/3
1
0
V22V2/3
d > 50%0
V2/2
V2/30
-2V2/3-V2/6
1
0
vcg*
d = 50% d = 50%
v0c
d < 50% d < 50%
d > 50% d > 50%
S21
S22
S23
S24
d > 50% d = 50%
vcg*
vcg*
vdg*
vdg*
vcg
No OCFin 1st arm
NNo OCFin 2nd arm
Waveforms transition
S22
S23
vcd
D22
D23
S21
S24
D26
D27
C2
C3
V2
i2
D21
D24
D25
D28
D1
D2
D3
D4
c
d
+
o
g (a) (b) Fig. 8. Equivalent structures of the MV side with the SBA method when: (a) the faulty switch is in the first arm. (b) the faulty switch is in the second arm.
t
DThs
t0
vcd
iL
Ths
vab
tDThst0
vcd
iL
Ths
vab
(a) (b) Fig. 9. Steady-state waveforms with the SBA method. (a) Mode 1: in high power range. (b) Mode 2: in low power range.
t
OCF on S21
vab’
vcd
iL
t
Complementary interval
DThs DThs
vab’
vcdiL
t
Symmetric waveforms
vab’
vcdiL
DThs
Fig. 10. Complementary faulty interval for the OCF condition on S21.
vcd
D22
D23
S26
S27
D26
D27
S25
S28
C2
C3
V2
i2
D21
D24
D25
D28
D1
D2
D3
D4
c
d
+
o
g
S22
S23
S24
S26
S27
S25
S28
C2
C3
V2
i2
D1
D2
D3
D4
c
d
+
o
g
S21
ioiL
(a) (b) Fig. 11. Current conduction paths during [t0Ths, DThs] when blocking the gate-driving signal of: (a) S24. (b) S25.
S22
S23
S24
S26
S27
S25
S28
C2
C3
V2
i2
D1
D2
D3
D4
c
d
+
o
g
S21
ioiL
low power range, respectively, which are defined as Mode 3 and Mode 4. The expressions of the zero-crossing point t0, the transferred power, and the power range can be calculated as those in the SBA method, and the transferred power range can be obtained as
2 2 20
20
2( 2 1) 3 (16 24 5) (4 3) ,Mode 3
0 2( 2 1) 3, Mode 4
k k P k k k
P k k
(6) The power-transfer capacity of the SBA and the proposed
CSB method is shown in Fig. 14, where the upper boundaries of the power ranges for the two methods are illustrated. It can be seen that with variable values of k, the maximum power with the proposed CSB method is higher than that of the SBA method, which means the converters increase the power transmission capacity by employing the CSB method, especially when k is much lower than unity.
IV. SIMULATION RESULTS To verify the performances of the proposed fault-tolerant
control method, simulation results are provided. The main parameters are: the input voltage is 200 V, the transformer turns ratio n is 1, the auxiliary inductor is 100 μH, the capacitors C1, C2, and C3 are 1 mF, and the switching frequency is 10 kHz.
Fig. 15 shows the waveforms when the OCF occurs on S21, where the reference output voltage is 300 V (k = 0.67),
and the transferred power is 4500 W. It can be seen that under the OCF state, the waveform of vcd becomes asymmetric, where the DC bias occurs and the current stress increases significantly. Fig. 16 shows the simulation waveforms of the two capacitor voltages VC2 and VC3. It can be seen that during the faulty state, the value of VC2 increases, because the current io draws from the midpoint o, and upper capacitor charges during the faulty interval, resulting in the capacitor voltage imbalance.
Fig. 17 shows the waveforms with the SBA method. It should be noted that the system is controlled by a closed-loop control scheme. From Fig. 17, it can be seen that when the SBA method is employed, the DC bias and the overshoot current can be eliminated. However, the output voltage
S22
S23
vcd
D22
D23
S21
S24
D26
D27
C2
C3
V2
i2
D21
D24
D25
D28
D1
D2
D3
D4
c
d
+
o
g
S26
S27
(a) (b) Fig. 12. Equivalent structures of the MV side with the CSB method when: (a) the faulty switch is S21 or S24. (b) the faulty switch is S25 or S28.
t
DThs
t0
vcd
iL
Ths
vab
tt0DThs
vcd iL
Ths
vab
(a) (b) Fig. 13. Steady-state waveforms with the CSB method. (a) Mode 3: in high power range. (b) Mode 4: in low power range.
0.5 0.6 0.7 0.8 0.9 10
0.2
0.4
0.6
0.8
1
CSB method
SBA method
Voltage conversion ratio k
Uni
fied
pow
erra
nge
Fig. 14. Comparative curves of the upper boundaries of the power range between the SBA and CSB methods.
vcd
D22
D23
S26
S27
D26
D27
S25
S28
C2
C3
V2
i2
D21
D24
D25
D28
D1
D2
D3
D4
c
d
+
o
g
S22
S23
150200250300
-400
40
Vol
tage
s(10
0V/d
iv)
Cur
rent
(20A
/div
)
0.1 ms/divOCF state
V2 (V)5 ms/div
iL (A)
vab vcd
iL
Normal state
Fig. 15. Simulation waveforms when the OCF occurs on S21.
0
100
200
300
vC3 (V)
vC2 (V)
50 ms/div
OCF occurs
Fig. 16. Simulation waveforms of the capacitor voltages when the OCF occurs on S21.
150200250300
-400
40
Output voltage offset5 ms/div
Vol
tage
s(10
0V/d
iv)
Cur
rent
(20A
/div
)vab vcd
iL
0.1 ms/divNormal state SBA method
V2 (V)
iL (A)
Fig. 17. Simulation waveforms with the SBA method.
decreases significantly, which means the reference voltage and power cannot be achieved by using the SBA method, even with the largest phase-shift ratio D regulating by PI controller.
Fig. 18 shows the waveforms with the proposed CSB method. From Fig. 18, it can be seen that by using the CSB method, the DC bias and overshoot current are eliminated. Therefore, the CSB method can bring the 2/3-level DAB converter back to the safe operation. Furthermore, according to the waveform of the output voltage V2, it can be seen that the reference output voltage can be achieved by the closed-loop control, which means the power range is wider than that with the SBA method.
In addition, to verify that S24 is the complementary switch of S21 rather than S25, the waveforms of VC2 and VC3 when blocking the gate-driving signal of S24 and S25 are shown in Fig. 19. From Fig. 19 (a), it can be seen that when OCF occurs on S21, and the gate-driving signal of S24 is blocked, the values of the capacitor voltages VC2 and VC3 remain equal during the fault-tolerant operation. However, when the gate-driving signal of S25 is blocked, from Fig. 19 (b), it can be seen that compared with Fig. 16, the capacitor voltage imbalancing condition will be exacerbated, which verify the theoretical analysis about the complementary-switch pairs.
V. CONCLUSION This paper proposed a fault diagnosis and fault-tolerant
control method for the 2/3-level DAB converters to address the open-circuit faults. The detailed analysis of OCFs on each power switch was discussed. Based on the analysis, a fault diagnosis method by using the midpoint voltage of each bridge arm is derived to identify the faulty power switch accurately. Furthermore, a fault-tolerant control method is proposed to overcome the OCF effects such as DC bias, overshoot current and capacitor voltage imbalance. In this method, every two power switches form a complementary pair. When one of them is detected to be faulty, the gate signal of another one will also be blocked. Simulation results have verified that the DC bias and overshoot current can be reduced with the proposed CSB method, and the power transfer capability is increased compared with the traditional fault-tolerant method, improving the reliability performance of the 2/3-level DAB converters.
REFERENCES [1] M. S. Mohammed and R. A. Vural, “Evolutionary Design Automation
of High Efficiency Series Resonant Converter for Photovoltaic Systems,” IEEE Trans. Power Electron., vol. 35, no. 11, pp. 11332-11343, Nov. 2020.
[2] Q. Tian, G. Zhou, M. Leng, G. Xu, and X. Fan, “A Nonisolated Symmetric Bipolar Output Four-Port Converter Interfacing PV-Battery System,” IEEE Trans. Power Electron., vol. 35, no. 11, pp. 11731-11744, Nov. 2020.
[3] X. Li, M. Zhu, M. Su, J. Ma, Y. Li, and X. Cai, “Input-Independent and Output-Series Connected Modular DC–DC Converter With Intermodule Power Balancing Units for MVdc Integration of Distributed PV,” IEEE Trans. Power Electron., vol. 35, no. 2, pp. 1622-1636, Feb. 2020.
[4] M. A. Moonem and H. Krishnaswami, “Analysis and Control of Multi- Level Dual Active Bridge DC-DC Converter,” in Proc. ECCE, Raleigh, 2012, pp. 1556-1561.
[5] S. Yang, A. Bryant, P. Mawby, D. Xiang, L. Ran, and P. Tavner, “An Industry-Based Survey of Reliability in Power Electronic Converters,” IEEE Trans. Ind. Appl., vol. 47, no. 3, pp. 1441-1451, May-June 2011.
[6] N. Zhao, J. Liu, Y. Shi, J. Yang, J. Zhang and X. You, “Mode Analysis and Fault-Tolerant Method of Open-Circuit Fault for a Dual Active-Bridge DC-DC Converter,” IEEE Trans. Ind. Electron., vol. 67, no. 8, pp. 6916-6926, Aug. 2020.
[7] D. Xie and X. Ge, “Open-Circuit Fault Diagnosis of Dual Active Bridge DC-DC Converter Based on Residual Analysis,” in Proc. PEAC, Shenzhen, 2018, pp. 1-6.
[8] B. Shi, B. Zhou, Y. Zhu, X. Qin, J. Lei, and N. Han, “Open-Circuit Fault Analysis and Diagnosis for Indirect Matrix Converter,” IEEE J. Emerg. Sel. Topics Power Electron., vol. 6, no. 2, pp. 770-781, Jun. 2018.
[9] A. Davoodi, N. Noroozi, and M. R. Zolghadri, “A Fault-Tolerant Strategy for Three-Phase Dual Active Bridge Converter,” in Proc. PEDSTC, Shiraz, Iran, 2019, pp. 253-258.
[10] M. Zheng, H. Wen, H. Shi, Y. Hu, Y. Yang, and Y. Wang, “Open-Circuit Fault Diagnosis of Dual Active Bridge DC-DC Converter With Extended- Phase-Shift Control,” IEEE Access., vol. 7, pp. 23752-23765, 2019.
[11] H. Wen, J. Li, H. Shi, Y. Hu, and Y. Yang, “Fault Diagnosis and Tolerant Control of Dual-Active-Bridge Converter with Triple-Phase-Shift Control for Bidirectional EV Charging Systems,” IEEE Trans. Transp. Electrific., vol. 7, no. 1, pp. 287-303, Mar. 2021.
[12] H. Shi, H. Wen, and J. Li, “Fault Analysis and Fault-Tolerant Method of Dual Active Bridge Converter under Triple Phase Shift Control,” in Proc. ECCE Asia, Busan, Korea, 2019, pp. 1121-1127.
150200250300
-400
40
5 ms/div
Vol
tage
s(10
0V/d
iv)
Cur
rent
(20A
/div
)
0.1 ms/div
vcdvab
iLNormal state CSB control
V2 (V)
iL (A)
Fig. 18. Simulation waveforms with the proposed CSB method.
0
100
200
300
50 ms/div
Block S24
vC3 (V)
vC2 (V)
(a)
0
100
200
300
50 ms/div
Block S25
vC3 (V)
vC2 (V)
(b)
Fig. 19. Simulation waveforms of the capacitor voltages under faulty state when blocking the gate-driving signal of: (a) S24. (b) S25.