This article has been accepted for publication in a future issue of this journal, but has not been

fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIE.2018.2793260, IEEE
Transactions on Industrial Electronics

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS

Rectifier Load Analysis for Electric Vehicle

Wireless Charging System
Yanjie Guo, Lifang Wang, Member, IEEE, Yuwang Zhang,
Shufan Li, and Chenglin Liao, Member, IEEE

 L2 R2 Z2 Self-inductance, resistance, and self-impedance

Abstract—This paper presents the analysis of rectifier of the receive coil.
load used for electric vehicle (EV) wireless charging M Zm Mutual-inductance and mutual-impedance
system, as well as its applications on compensation between the two coupling coils.
network design and system load estimation. Firstly, a Lp RLp Zp Inductance, resistance, and impedance of the
rectifier load model is established to get its equivalent primary side compensation inductor.
input impedance, which contains both resistance and
inductance components, and can be independently
Ls RLs Zs Inductance, resistance, and impedance of the
calculated through the parameters of the rectifier circuit. secondary side compensation inductor.
Then, a compensation network design method is proposed, C1s R1s Z1s Capacitance, resistance, and impedance of the
based on the rectifier load analysis. Furthermore, a primary side series compensation capacitor.
secondary side load estimation method and a primary side C1p R1p Z1p Capacitance, resistance, and impedance of the
load estimation method are put forward, which adopt only primary side parallel compensation capacitor.
measured voltages and consider the influence of the C2s R2s Z2s Capacitance, resistance, and impedance of the
rectifier load. Finally, an EV wireless charging prototype is secondary side series compensation capacitor.
developed, and experimental results have proved that the C2p R2p Z2p Capacitance, resistance, and impedance of the
rectifier equivalent load can be correctly calculated on
conditions of different system load resistances, rectifier
secondary side parallel compensation capacitor.
input inductances, DC voltages, and mutual-inductances. Cin System input filter capacitor.
The experiments also show that rectifier load equivalent Co RCo Capacitance and resistance of the system output
inductance will impact system performances, and the filter capacitor.
proposed methods have good accuracy and robustness in RL System equivalent load resistance.
the cases of system parameter variations. ω System angle frequency.
us Vs Voltage source of the rectifier circuit and its
Index Terms—Wireless charging system, rectifier load, amplitude.
compensation network design, load estimation. urec irec Rectifier input voltage and current.
θb θf Start and end phase angles of urec and irec.
udio Vdio Diode forward voltage drop and its value.
NOMENCLATURE Rdio Diode conduction resistance.
Ud DC voltage source. ud Vd Load voltage and its approximate DC value.
G1 - G4 Primary side MOSFETs. id Id Load current and its approximate DC value.
D1 - D4 Secondary side diodes. Φ(t) Characteristic matrix of the rectifier circuit.
L1 R1 Z1 Self-inductance, resistance, and self-impedance Urec_fd φurec_fd Fundamental amplitude and phase angle of urec.
of the transmit coil. Irec_fd φirec_fd Fundamental amplitude and phase angle of irec.
Re Le Series equivalent resistance and inductance of
Manuscript received May 11, 2017; revised August 29, 2017 and the rectifier load.
November 30, 2017; accepted December 21, 2017. This work was Zs1 Impedance after the secondary side series
supported by National Natural Science Foundation of China (51507168), compensation capacitor C2s.
International Science & Technology Cooperation Program of China Zp1 Impedance after the primary side compensation
(2016YFE0102200), and the State Key Laboratory of Automotive Safety
and Energy of China (KF16012). inductor Lp.
Y. J. Guo, L. F. Wang, and C. L. Liao are with the Key Laboratory of ηc Efficiency from inverter output to rectifier load
Power Electronics and Electric Drives, Institute of Electrical Engineering, equivalent impedance Re and Le.
Chinese Academy of Sciences, Beijing 100190, China, and also with the Ropt Optimal resistance in the secondary side.
Beijing Co-Innovation Center for Electric Vehicles, Beijing 100081,
China (e-mail: yjguo@mail.iee.ac.cn; wlf@mail.iee.ac.cn; liaocl@mail. Res Equivalent resistance of the secondary side,
iee.ac.cn). when C2s and C2p are well designed.
Y. W. Zhang, and S. F. Li are with the Key Laboratory of Power uinv Uinv Inverter output voltage and its RMS value.
Electronics and Electric Drives, Institute of Electrical Engineering, Por ηr Rated system output power and efficiency.
Chinese Academy of Sciences, Beijing 100190, China, and also with the
University of Chinese Academy of Sciences, Beijing 100190, China Lsoft Inductance needed for inverter soft switching.
(e-mail: zhangyuwang@mail.iee.ac.cn; lishufan@mail.iee.ac.cn). RL_Sesti Estimated load resistance adopting the proposed

0278-0046 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIE.2018.2793260, IEEE
Transactions on Industrial Electronics

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS

secondary side load estimation method. domain, or complex frequency domain [13,14], which can be
RL_Pesti Estimated load resistance adopting the proposed used for the analysis of WCS rectifier equivalent load
primary side load estimation method. impedance. Besides, non-linear switching functions and circuit
tucs Positive zero crossing time of the voltage before simulations could also be adopted to study this issue [15].
rectifier input inductor. The non-linear process of rectifier load will bring some
turs Positive zero crossing time of the rectifier input difficulties to system compensation network design. As we
voltage following tucs. know, compensation networks are very important to system
Gp Fundamental voltage transfer function between performances [16], and can be designed to achieve maximum
the inverter output voltage and the voltage after efficiency, maximum power, or conjugate matching [17,18]. In
inverter output inductor.
most cases, a pure resistance is used to express the rectifier load
Gs Fundamental voltage transfer function between
[19-21]. But the operation modes of WCS rectifier load will
the voltage before rectifier input inductor and
affect the working states of compensation network [22]. So,
the rectifier input voltage.
Z11 Z12 Z21 Z22 Impedance parameters of the equivalent actual equivalent input impedance of WCS rectifier load should
two-port network of the coupling coils and be considered, while designing the compensation networks.
compensation capacitors. Load estimation of WCS has faced the same problem.
n1 n2 d1 d2 Impedance coefficients in the relationship Effects of the rectifier load could complicate the equations used
equation between Gp and Gs. for load estimation [23], and lead to the increasing of
Amp Php Amplitude and phase angle of Gp. calculation and control complexity. Hence, a pure resistance
θn θd Phase angles of the numerator and denominator load is approximately used for most of the load estimation,
of Gs. detection, or optimal load tracking [24-26]. Another situation is
amn1 phn1 Amplitude and phase angle of n1. that the voltages and currents are usually both measured for
amd1 phd1 Amplitude and phase angle of d1. load estimation, in order to calculate the impedances in the
ren2 imn2 Real and imaginary parts of n2. primary side [24,27]. Since the voltage and current sensors or
red2 imd2 Real and imaginary parts of d2. probes have different phase delays at the high frequency range,
Tj Diode junction temperature in degrees Celsius. some deviations may be introduced into the estimation process.
Also, the robustness of the estimation method is very important.
I. INTRODUCTION It can be analyzed through parameter derivation, root locus,

E LECTRIC vehicle (EV) wireless charging system (WCS)

has the advantages of convenience, space-saving, etc. So, it
has attracted much attention. In recent years, working principle,
Nyquist curve, Bode graph, or directly calculating the results on
conditions of parameter variations [28-30].
Based on the previous researches, an effective method to
operation characteristics, system design, and control method of quantitatively analyze the equivalent load of WCS rectifier is
both stationary and dynamic wireless EV charging systems put forward in the paper firstly. The equivalent load can be
have been studied and applied to some demonstrations [1,2]. independently calculated through the parameters of the rectifier
In applications of EV wireless charging, rectifier and output circuit, and the results are basically not affected by other WCS
filter capacitor are needed to convert the high frequency AC to parts. Secondly, a compensation network design method is
DC, in order to charge the power battery. Rectifier and the proposed considering the equivalent impedance of the rectifier
circuit after it are usually equivalent to a pure resistance load to load, especially the equivalent inductance. This method will
design the system or control strategy [3,4]. A conventional way further decouple the primary and secondary side design, to
is using the coefficient 8/π2 to make an equivalent relationship achieve four system performance indicators at the same time.
between the rectifier input impedance and the system load Thirdly, the effects of the rectifier non-linear process are taken
resistance [5,6]. However, stray parameters and non-ideal into count to estimate the system load resistance. The proposed
behaviors of the devices will become obvious at the high primary side load estimation method only adopts high
frequency range [7]. Also, rectifier input impedance can be frequency voltages, does not need to measure the currents, and
affected by the input inductance and other parameters. So, it can avoid the phase delay deviations. Also, it does not require
will bring some deviations, if only considering WCS rectifier wireless communication between the primary and secondary
input impedance as a pure resistance. sides. This paper is organized as follows. Section II establishes
Actually, rectifier input impedance of EV wireless charging the rectifier load model and shows the equivalent input
system contains both resistance part and inductance part [7-9]. impedance calculations. Section III presents the compensation
It can be expressed as a series of an equivalent resistance and an network design method. Section IV proposed the load
equivalent inductance [8,9]. Although there has not been an estimation methods based on primary and secondary side
effective method to get the equivalent load impedance of WCS voltage measurements. Section V gives the experimental
rectifier, some existing researches could be helpful. Based on verifications and discussions.
the on and off states [10], the rectifier and its related inductance
and capacitance circuits can be described by the state space II. RECTIFIER LOAD ANALYSIS AND CALCULATION
model [11], considering the stray resistances and diode forward Full-bridge diode rectifier is the most commonly used
voltage drop [12]. Then, the expressions of the related voltages topology in EV wireless charging system. Also, dual-side LCC
and currents have been obtained in the time domain, frequency

0278-0046 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIE.2018.2793260, IEEE
Transactions on Industrial Electronics

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS

compensation networks can provide several appropriate design avoid too large current peaks in the diodes. Hence, only CCM
degrees of freedom to achieve several system performance states are shown in Fig.2, and discussed in this paper. Besides,
indicators at the same time. Moreover, it can be designed to the steady state waveforms of urec and irec are presented in Fig.2,
make the system resonant frequency independent of the load when only a few fluctuations exist on the voltage of the output
condition [16,22]. So we discuss the rectifier load on the basis capacitor Co and the voltage drop on RCo is very small. So, urec
of this kind of topology. can be approximately described as a square wave.
Fig.2 suggests that the waveform of rectifier input current irec
has some distortion, because of the effect of the rectifier input
inductance. This makes the fundamental wave of irec lags
behind the one of urec. So, the rectifier input impedance does not
just include resistance component, but also contains a certain
inductance component. Moreover, Fig.2 shows that the positive
Fig. 1. EV wireless charging system with full-bridge diode rectifier and and negative half-cycles are symmetric for all the voltage and
dual-side LCC compensation networks. current waveforms. Hence, we just need to consider the positive
Fig.1 shows the EV wireless charging system with half-cycle, and the negative half-cycle can be obtained from the
full-bridge diode rectifier and dual-side LCC compensation symmetry. Fig.3 shows the equivalent circuit of the rectifier
networks; where, Ud is DC voltage source; the high frequency circuit in the positive half cycle, considering the stray
inverter is composed of G1-G4, and the full-bridge rectifier is parameters and the diode forward voltage drop; where, udio
composed of D1-D4; the primary side compensation network represents the diode forward voltage drop; Rdio is diode
consists of Lp, C1s, and C1p; the secondary side compensation conduction resistance; RLs and RCo are stray resistances of Ls
network consists of Ls, C2s, and C2p; L1 and L2 are and Co, respectively; ud and id are load voltage and current.
self-inductances of the transmit coil and receive coil; M is
mutual-inductance between them; Cin and Co are system input
and output filter capacitors; RL is system load resistor. It should
be noticed that the WCS load is an EV power battery in the
practical case, which behaves as a voltage source series with its
parasitic resistance. But the power battery could be equivalent
to a load resistance RL [1,19]; the value of this equivalent
resistance can be calculated by the voltage on the power battery
divided by the current flowing through it. Moreover, the Fig. 3. Equivalent circuit of the rectifier circuit in the positive half cycle.
full-bridge rectifier, its input inductor, output filter capacitor, Based on the equivalent circuit, irec is defined as state
and the load resistor are together defined as the rectifier circuit. variable x1, and the voltage on Co is defined as state variable x2.
Although the following analysis is conducted based on the us and udio are treated as the input variables, and ud is treated as
specific system, it can be extended to applications on other the output variable. So, state space equation of the rectifier
rectifier and compensation network topologies. circuit in the positive half cycle is given by (1a).
x   x   u  x 
 1   A  1   B  s  , y  C  1  (1a)
 x2   x2  udio   x2 
Where, impedance matrixes A, B, and C are given by (1b).
 1 RL RCo 1 RCo 
  L ( RLs  2 Rdio  R  R ) L (1  R  R ) 
A ,
s L Co s L Co

 RL 1  (1b)
   
 Co ( RL  R Co ) Co ( R L  RCo ) 
Fig. 2. Schematic waveforms of the source voltage, rectifier input
1/ Ls  2 / Ls   RL RCo RCo 
voltage and current. B  ,C  R  R 1  R  R   
In order to calculate rectifier equivalent input impedance, we  00   L Co L Co 

firstly need to analyze the voltages and currents of rectifier Then, the input variables and the initial values of the state
circuit, which are shown in Fig.2; where, us is the voltage on variables are given by (2), according to the schematic
C2p, which is a sine wave [22], and can be treated as the voltage waveforms in Fig.2; where, ω is system angle frequency; the
source of the rectifier circuit; urec and irec are rectifier input diode forward voltage drop is treated as a constant value Vdio.
voltage and current; the start time of us positive half-cycle is Since only a few fluctuations exist on the voltage of Co and the
selected as the coordinate zero of x-axis. θb and θf are start and voltage drop on RCo is very small, their influences can be
end phase angles of urec and irec. So, θf =θb+π. Also, the rectifier ignored, and the initial value of x2 can be approximately
input inductance Ls should be big enough to keep the rectifier equivalent to a DC voltage variable Vd. Also, amplitude of us is
working in the continuous conduction mode (CCM), in order to defined as Vs, and it will be affected by WCS parameters, such

0278-0046 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIE.2018.2793260, IEEE
Transactions on Industrial Electronics

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS

as source voltage, mutual-inductance, etc. But the amplitudes Finally, the fundamental wave amplitudes and phase angles
of urec and irec are proportional to Vs. So, Vs can be treated as a of urec and irec can be calculated through Fourier transform, and
known variable. defined as Urec_fd, Irec_fd, φurec_fd, and φirec_fd. So, the equivalent
us   Vs sin(t   b ),udio  Vdio , x (0)   0, Vd  . (2) input impedance of WCS rectifier load will be given by (9);
T

where, Re and Le are series equivalent resistance and inductance

Furthermore, Vd and θb should be calculated to solve the state of the rectifier load. Only fundamental wave is considered,
space equation. On the WCS normal working conditions, the because the power of the harmonics is much smaller than the
value of Vdio and the voltage drops on Rdio and RLs are much one of the fundamental wave. But the harmonic input
smaller than the ones of Vs and Vd. So, the voltage on Ls is impedances can also be obtained from Fourier transform.
approximately equivalent to Vs sinθ - Vd, and the expression of Moreover, the calculation process suggests Re and Le will be
irec can be given by (3), according to the relationship between affected by the parameters of the rectifier circuit. Hence, the
the voltage on an inductor and the current flowing through it. robustness of this method towards parameter variation needs to
1  be studied. But the theoretical methods, such as calculating the
irec 
 Ls  (V sin   V )d .
b
s d
(3)
derivative and root locus, cannot provide a simple and clear
way to analyze the robustness in this case, since it is related to
As shown in Fig.2, irec=0, when θ=θf=θb+π. So, one some complex or non-linear operations. So, this issue will be
relationship between Vd and θb can be got and given by (4). discussed in Section V, based on the actual parameter values.
Vd  (2Vs cos  b ) /  . (4) Re  (U rec _ fd / I rec _ fd ) cos(urec _ fd  irec _ fd ),
Also, the DC load current Id can be calculated by (5), which (9)
is the average value of id in the positive half cycle.
Le  (U rec _ fd / I rec _ fd ) sin(urec _ fd  irec _ fd ) / .
1  b   To sum up, the above analysis suggests that the rectifier load
Id 
 Ls  
b b
(Vs sin   Vd )d
(5) equivalent impedance contains both resistance and inductance
components. Also, the series equivalent resistance and
 (Vs (2sin  b   cos b )   2Vd / 2) /  Ls . inductance can be independently calculated through parameters
Because Id =Vd /RL, another relationship between Vd and θb of rectifier circuit, and the results are basically not affected by
can be got and given by (6). other WCS parameters. So, the rectifier load can be decoupled
with other parts of WCS, and make system design easier.
Vd  Vs (2 sin  b   cos  b ) / ( ( Ls / RL   / 2)). (6)
Based on the two relationships between Vd and θb, they can III. COMPENSATION NETWORK DESIGN
be obtained from (4) and (6). The expression of θb is given by Since the rectifier load has been decoupled with other parts
(7), and the expression of Vd can also be got according to their of WCS, we are going to propose a compensation network
relationships. Equation (7) indicates that the phase difference design method, based on the rectifier load analysis and some
between us and urec (or irec) is mainly decided by Ls and RL, and existing researches [16-18]. Moreover, the proposed method
approximately independent of other WCS parameters. Since will further decouple the primary and secondary side design,
amplitudes of urec and irec are basically proportional to the one and make the WCS compensation network design simpler. As
of us as mentioned above, we can say that the other parts of same as the rectifier load analysis, the dual-side LCC
WCS have little effect on the rectifier circuit, and the rectifier compensation networks are used here. The rectifier input
load can be decoupled to analyze its equivalent input inductance Ls should be big enough to keep the rectifier
impedance. It is should be noticed that the rectifier circuit working in CCM state as mentioned above, so we will confirm
seems to be equivalent to a pure resistance RL, according to (7). it before the compensation network design. Also, the primary
However, this equivalent relationship is only suitable for (7) side compensation inductance Lp is assumed to be known, and
when calculating the phase angle θb, and cannot be used for any only the four compensation capacitors are used in the design
other part in the rectifier load analysis. method in this section.
b  arctan( Ls / RL ). (7)
After getting Vd and θb, full response of the rectifier circuit in
the positive half cycle can be calculated by (8); where, Φ(t) is
the characteristic matrix of rectifier circuit; the part before the
plus sign is used for solving zero-input response, and the other
part is used for solving zero-state response. On the basis of (8),
time domain expressions of urec and irec can be obtained, Fig. 4. Equivalent circuit of the system secondary side, considering
rectifier load equivalent impedance.
according to the symmetry of their waveforms.
t Firstly, the secondary side is discussed, and its equivalent
x (t )   (t ) x (0)    ( )Bu(t   )d circuit is shown in Fig.4; where, the series equivalent resistance
0
(8) Re and equivalent inductance Le are used to express the rectifier
 0  t Vs sin( (t   )  b ) 
 e     e A B 
At
d .
load; R2 is resistance of the receive coil. As shown in Fig.4, Zs1
Vd  0
 Vdio  is defined as the impedance after the secondary side series

0278-0046 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIE.2018.2793260, IEEE
Transactions on Industrial Electronics

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS

compensation capacitor C2s, and its expression is given by (10); given by (14); where, Xpe=ωL1-1/(ωC1s)-1/(ωC1p); re(Zp1)
where, Re’=Re+RLs; Le’=Le+Ls; re(Zs1) means the real part of Zs1; means the real part of Zp1; im(Zp1) is the imaginary part of Zp1.
im(Zs1) is the imaginary part of Zs1. Similar with the secondary side design, the primary side also
contains two compensation capacitors with two degrees of
Re / ( 2C2 p 2 )
re( Z s1 )  , freedom for design. So, two design targets could be added here.
Re2  ( Le  1/ (C2 p )) 2 (10)
The first one is making the WCS output rated power. The
corresponding target equation is given by (15); where, Uinv is
Re / (C2 p )  Le ( Le  1/ (C2 p )) / C2 p
2
the RMS value of uinv; Por is the rated WCS output power; ηr is
im( Z s1 )   .
Re2  ( Le  1/ (C2 p )) 2 the rated WCS efficiency.

So, expression of the efficiency ηc can be calculated and U inv 2 / re( Z p1 )  Por / r . (15)
given by (11); where, ηc is the efficiency from inverter output The second design target is keeping the input impedance of
to rectifier load impedance; R1 is resistance of the transmit coil; the primary side compensation network containing a certain
Xse=im(Zs1)+ωL2-1/(ωC2s). inductance, in order to realize the soft switching of the inverter.
re( Z s1 ) 2 M 2 The corresponding target equation is given by (16); where, Lsoft
c  . (11) is the inductance needed for inverter soft switching.
(re( Z s1 )  R2 ) M 2  (re( Z s1 )  R2 ) 2 R1  R1 X se 2
2

im( Z p1 ) /   Lp  Lsoft . (16)

Equation (11) indicates that two conditions need to be met,
for the sake of achieving maximum efficiency. One is Xse=0 to Through simultaneously solving (15) and (16), values of the
minimize the denominator of ηc. The other is the load resistance primary side compensation capacitors C1s and C1p can be
of the receive coil is equal to the optimal load resistance Ropt, as obtained, which is not affected by the secondary side design
given by (12); where, Ropt are obtained from the derivation of ηc, process. Also, it should be noticed that sometimes there is no
when Xse=0. analytical solution for these equations. Numerical solution
methods need to be used on this condition.
re( Z s1 )  Ropt  R2 2   2 M 2 R2 / R1 . (12) Finally, the primary and secondary side compensation
On the basis of (10) and (12), the secondary side parallel networks have been decoupled for design. Also, four
compensation capacitor C2p can be calculated and given by (13). compensation capacitors with four degrees of freedom are
According to the value of C2p and the equation Xse=0, the designed by considering four system performance indicators,
secondary side series compensation capacitor C2s can also be including achieving maximum efficiency, optimal load
solved. The above analysis suggests that the secondary side resistance, making WCS output rated power, and realizing the
compensation capacitors can be designed independently of the soft switching of the inverter. Besides, calculated values of the
primary side ones, and their design purpose is mainly to achieve designed compensation capacitors require fine tuning in
maximum system efficiency. practice to get better results.

 Le   2 Le2  ( Re2   2 Le2 )(1  Re / Ropt ) IV. LOAD ESTIMATION METHODS
C2 p  . (13)
( Re2   2 Le2 ) The rectifier load analysis results can be used for system load
estimation, which adopting the high frequency signals in WCS.
Then, the primary side is studied, and its equivalent circuit is The conventional load estimation methods are usually based on
shown in Fig.5; where, uinv is inverter output equivalent voltage the pure resistance load, and also need the high frequency
source; RLp is stray resistances of Lp; Res is the equivalent voltage and current at the same time [24,27]. The voltage and
resistance of the secondary side, when C2s and C2p are well current sensors or probes will have different phase delays at the
designed, and Res= ω2M2 /(Ropt+R2) . high frequency range, including the ones used in oscilloscopes
and power analyzers. These different phase delays will lead to
some deviations of the phase angle between the measured
voltage and current, and affect the accuracy of the impedance
calculation, especially when the phase angle is close to 90°.
In order to solve this problem, we propose a load estimation
method based on the secondary side high frequency voltages.
Fig. 5. Equivalent circuit of the system primary side, when the
secondary side is well designed. The specific process is as follows: firstly, the positive zero
crossings of the rectifier input voltage (urec) and the voltage
( Res  R1 ) / ( 2C1 p 2 ) before rectifier input inductor (the voltage on C2p for LCC
re( Z p1 )  ,
( Res  R1 ) 2  X pe 2 topology) are detected, in order to obtain the positive zero
(14)
crossing times. Then, define the positive zero crossing time of
( Res  R1 )  X pe ( L1  1/ (C1s ))
2
the voltage before rectifier input inductor as tucs, and the
im( Z p1 )   .
C1 p (( Res  R1 ) 2  X pe 2 ) following positive zero crossing time of the rectifier input
As shown in Fig.5, Zp1 is defined as the impedance after the voltage as turs. So, the load estimation expression is given by
primary side compensation inductor Lp, and its expression is (17), according to the relationship shown in (7). Finally, since

0278-0046 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIE.2018.2793260, IEEE
Transactions on Industrial Electronics

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS

WCS has been built before load estimation, the value of the Where, the coefficients n1, n2, d1, and d2 are defined by (19b);
rectifier input inductor Ls can be measured, and system angle where, Zp=RLp+jωLp; Zs=RLs+jωLs.
frequency ω is also known. So, the estimated load RL_Sesti can be n1  Z12 Z 21  Z11Z 22 d 2   Z11  Z p   Z 22  Z s   Z12 Z 21 
calculated through (17). (19b)
RL _ Sesti   Ls / tan((turs  tucs )). (17) n2  Z p Z 22  Z11Z 22  Z12 Z 21 d1  Z12 Z 21  Z11  Z 22  Z s  .
The proposed secondary side load estimation method has Furthermore, the amplitudes and phase angles of the selected
considered the influence of the WCS rectifier load. Also, only voltages will be measured in the primary side, and then the
high frequency voltages are used in this method; no current is transfer function Gp can be obtained. Define the amplitude of
adopted. Hence, it can avoid the deviations introduced by Gp as Amp, and the phase angle of Gp as Php. So, θn, which is
different phase delays between measured voltage and current. the phase angle of the numerator of Gs, can be calculated, as
Besides, the proposed method only detects the positive zero well as θd, which is the phase angle of the denominator of Gs.
crossing times, but does not need the voltage amplitudes or Their expressions are given by (20).
RMS values. This will bring some simplifications to the Amp  amn1 sin( Php  phn1)  imn 2
 n  arctan ,
corresponding measurements and calculations. Amp  amn1 cos( Php  phn1)  ren2 (20)
However, the measured signals still need to be transmitted to Amp  amd1 sin( Php  phd1)  imd 2
the primary side by wireless communication in most cases,  d  arctan .
used for system optimization or control. In order to avoid the
Amp  amd1 cos( Php  phd1)  red 2
problems brought by wireless communication, we further put Where, amn1 and phn1 are the amplitude and phase angle of n1;
forward a load estimation method based on the primary side amd1 and phd1 are the amplitude and phase angle of d1; ren2
high frequency voltages. Here, the inverter output voltage (uinv) and imn2 are the real and imaginary parts of n2; red2 and imd2
and the voltage after inverter output inductor (the voltage on are the real and imaginary parts of d2; they can be calculated
C1p for LCC topology) are adopted. Define the fundamental through (18) and (19), according to the measured values of the
voltage transfer function between the inverter output voltage WCS parameters.
and the voltage after inverter output inductor as Gp, and the Finally, we can get the phase angle of Gs, and the estimated
fundamental voltage transfer function between the voltage load RL_Pesti can be calculated through (21). Moreover, the
before rectifier input inductor and the rectifier input voltage as derivation process suggests that RL_Pesti will be affected by
Gs. The phase angle of Gs will be θb as defined in Fig.2, which WCS parameters, such as mutual-inductance M, compensation
can be adopted for load estimation based on (7). So, we need to capacitances C1s, C1p, C2s, C2p, and so on. Hence, the robustness
find a relationship between Gs and Gp, and then the measured of the estimation methods needs to be studied, when these
primary side voltages can be used to calculate θb. To achieve parameters vary. But similar with the case of the rectifier
this, some WCS parts can be treated as a two-port network equivalent load calculation method in Section II, the theoretical
[17,18]. Hence, the coupling coils and compensation capacitors methods cannot provide a simple and clear way to analyze the
are equivalent to a two-port network as shown in Fig.6. robustness. So, this issue will be also discussed in Section V,
based on the actual parameter values.
RL _ Pesti   Ls / tan( n   d ). (21)
Developed from the above secondary side load estimation
method, the proposed primary side load estimation method has
also considered the influence of the rectifier load. Meanwhile, it
only adopts high frequency voltages, and can avoid the phase
delay deviations, too. The difference is this method needs to
Fig. 6. Schematic of the equivalent two-port network and its parameters. measure voltage amplitudes. But on the other side, it does not
According to Fig.6, impedance parameters of the equivalent require wireless communication between the primary and
two-port network can be calculated and given by (18a). secondary sides. So, it has some advantages in EV applications.
Z11  Z1 p (( Z1  Z1s )( Z 2  Z 2 s  Z 2 p )  Z m 2 ) / den, V. EXPERIMENTAL VERIFICATION AND ANALYSIS
Z12  Z 21  Z1 p Z 2 p Z m / den, (18a)
A. Experiment Layout and System Parameters
Z 22  Z 2 p (( Z 2  Z 2 s )( Z1  Z1s  Z1 p )  Z m ) / den.2
An EV wireless charging prototype is developed to verify the
Where, Zm=jωM, and the denominator den is defined by (18b). rectifier load analysis results and the proposed methods. Its
configuration is shown in the photograph in Fig.7. A full-bridge
den  ( Z1  Z1s  Z1 p )( Z 2  Z 2 s  Z 2 p )  Z m 2 . (18b) single-phase inverter with MOSFETs is assigned as the power
Then, the relationship between Gs and Gp can be got and source. System load is a full-bridge diode rectifier with load
given by (19a), based on the impedance parameters of the resistors. The transmit coil is rounded rectangular with spiral
equivalent two-port network. disc type of fifteen turns of Litz wires. The receive coil is
square with spiral disc type of twenty turns of Litz wires. The
G s  ( n1 G p  n2 ) / ( d1 G p  d 2 ). (19a)
Litz wire used for coils is 640 strands with 0.1 mm diameter for

0278-0046 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIE.2018.2793260, IEEE
Transactions on Industrial Electronics

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS

each strand. Also, ferrites are adopted as the core material of fundamental waves will be calculated through the FFT (fast
the coils, and aluminum plates are added outside the ferrites for Fourier transform) program. So, the fundamental waves can be
more shielding. The transmit coil size is 58 cm  42 cm, and the drawn by Matlab and shown in Fig. 8(b). It suggests that the
receive coil size is 32 cm  32 cm. The vertical distance fundamental wave of rectifier input current lags behind the one
between the coils is 20 cm. of rectifier input voltage, which means the rectifier input
impedance contains a certain inductance component. This
conclusion is consistent with the one obtained in Section II.

(a) Rectifier input voltage and current waveforms.

Fig. 7. Photograph of the developed EV wireless charging prototype.

The prototype is designed with rated output power 3.3 kW on
the input DC voltage 400 V. System operation frequency is 85
kHz. Load resistance RL is 42.9 Ω, which is selected according
to an EV power battery with about 325 V - 340 V open-circuit
voltage and 8 A charging current. System impedance parameter
values are measured by a LCR meter, and the results are given
as follows. Self-inductances values of the transmit coil and (b) Calculated fundamental waves.
receive coil are 232.9 uH and 219.7 uH. Mutual-inductance Fig. 8. Experimental results of rectifier input voltage and current, as well
value is 25.4 uH, when the receive coil is aligned with the as their fundamental waves, based on the standard parameter values.
transmit coil. Compensation inductors Lp and Ls are selected to
be 79.9 uH and 83.3 uH. Based on the design method in Section
III, the compensation capacitors can be obtained as follows:
C1s=19.7 nF, C1p=82.8 nF, C2s=23.6 nF, C2p=69.6 nF. The
above parameter values are defined as the standard parameter
values, which can make the system achieve good performances,
such as rated output power, high efficiency, inverter soft
switching, etc. In the following sections, some of these
parameters will be changed to different values for further
verifications and discussions.
The type of MOSFET is IPW65R037C6. Also, the values of
Vdio and Rdio will vary with the junction temperature Tj (in
degrees Celsius), as shown in the datasheet of the diode
C3D16060D, which is used in the developed prototype. So,
their values can be obtained from the measured temperature by
a thermal imager and the following equations in the datasheet:
Vdio = 0.93+((-9.310-4)Tj); Rdio = 0.058+((5.710-4)Tj).
B. Rectifier Equivalent Load Verifications and Analysis
Based on the developed EV wireless charging prototype, the
calculation method of the rectifier load equivalent impedance is
verified, and the equivalent load characteristics are studied.
Firstly, experimental waveforms of the rectifier input voltage
and current are given in Fig. 8(a), under the condition of
standard parameter values. It shows that the waveform of
rectifier input current has some distortion, which is the same as
shown in Fig.2. Moreover, import the experimental data to the Fig. 9. Flow chart of the proposed method used to calculate the rectifier
software Matlab, and the amplitudes and phase angles of the equivalent load.

0278-0046 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIE.2018.2793260, IEEE
Transactions on Industrial Electronics

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS

Then, the rectifier equivalent load calculation method is Moreover, the calculated and experimental values of the
verified on conditions of different load resistances and rectifier equivalent resistances are close to the ones got by conventional
input inductances. Based on the analysis in Section II, the method (34.77 Ω and 17.43 Ω), which adopts the coefficient
equivalent resistance and inductance can be calculated through 8/π2 as the relationship between the rectifier input impedance
the process in the flow chart in Fig.9. and the system load resistance [5,6]. This indicates the
Following the steps in Fig.9, the equivalent impedances of conventional method can be approximately used for equivalent
the rectifier load can be calculated and given in Tab.1. Also, the resistance calculation. However, it ignores the equivalent
experimental ones are shown in Tab.1, which are obtained from inductance, as well as the effects of rectifier circuit parameters.
the fundamental amplitude and phase angle calculation results Furthermore, the influences of other WCS parameters on the
of the experimental waveforms. rectifier equivalent load are investigated. When DC source
TABLE I
voltage is changed to 200 V and other parameters still have
COMPARISONS BETWEEN CALCULATED AND EXPERIMENTAL RECTIFIER standard values, the calculated equivalent load is 33.1 Ω and
EQUIVALENT LOADS BASED ON DIFFERENT PARAMETER VALUES 10.5 uH; while, the experimental one is 33.1 Ω and 10.6 uH.
Prototype Calculated Experimental When the receive coil has a lateral misalignment distance of 10
Parameter Values Equivalent Load Equivalent Load cm, and other parameters still have standard values, the
RL (Ω) Ls (uH) Re (Ω) Le (uH) Re (Ω) Le (uH) calculated equivalent load is 33.1 Ω and 10.5 uH; while, the
83.3 33.1 10.5 33.0 10.2 experimental one is 32.9 Ω and 10.3 uH. These results suggest
42.9 113.9 33.8 7.7 33.7 8.2 other WCS parameters, such as DC source voltage and
49.1 30.4 16.9 32.7 18.2 mutual-inductance, basically have no effect on the rectifier
83.3 17.1 2.1 17.4 2.5 equivalent load. So, the rectifier load can be modeled and
21.5 113.9 17.3 1.4 17.2 1.7 analyzed, decoupling with other parts of WCS.
49.1 16.6 3.9 17.5 4.6 Finally, the robustness of the proposed rectifier equivalent
load calculation method is discussed towards parameter
Tab.1 suggests that the equivalent resistance and inductance uncertainties. The parameter uncertainties are mainly caused by
will both reduce when RL becomes smaller. While, the the measurement errors and the stray parameters. Also, the
equivalent resistance is not affected much by Ls, and the diode parameter values Vdio and Rdio will change with the
equivalent inductance will increase when Ls becomes smaller. junction temperature Tj. Since the junction temperature of the
Further calculations indicate that on the condition of standard diode cannot be measured directly, and only can be estimated
parameter values, the difference between the calculated and by the device surface temperature, so this will lead to parameter
experimental equivalent resistances is only 0.1 Ω, and the uncertainties of the rectifier, too. The nominal accuracy of the
calculation error of the equivalent inductance is smaller than LCR meter used for measurements is 0.05%. Hence, the
3%. This means the proposed method has good accuracy, when influence of the measurement error will be very small, and the
the system is well designed and working in the rated state. impact of Tj is adopted here to analyze the robustness.
However, when Ls is changed to 49.1 uH, the calculation error
has increased to 7%. This is caused by the value of Ls is too
small to make the rectifier working in the discontinuous
conduction mode (DCM). Also, when RL is changed to 21.5 Ω,
the relative error percentages increase, due to the small values
of equivalent inductances. But the absolute errors are still small.
So, the calculated results are basically close to the experimental
ones, which have proved the effectiveness of the rectifier load
model and the equivalent impedance calculation method.
However, there is a trend that Le becomes smaller when RL
(a) Calculation results of the equivalent resistance Re.
decreases. So, if RL has a small value about a few ohms, Le will
be very small. When the power battery voltage remains the
same while the charging current has increased to 80 A or higher,
the equivalent system load resistance will be a few ohms. This
case may happen when WCS is used for the fast charging of the
passenger cars, or the normal charging for the heavy duty
vehicles. The output power of this kind of WCS is usually more
than 20 kW, even reaches more than 100 kW. Under these
conditions, Le is much smaller than the rectifier input
inductance Ls, and its effect is ignorable. But the slow charging
(b) Calculation results of the equivalent inductance Le.
for the passenger cars (with output power lower than 10 kW)
Fig. 10. Rectifier equivalent load calculation results, when the diode
still hold a considerable proportion, so the proposed rectifier parameter values Vdio and Rdio change with the junction temperature Tj.
equivalent load calculation method could also be useful for the
The junction temperature Tj is usually 25-125 degrees
analysis and design of EV wireless charging system.

0278-0046 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIE.2018.2793260, IEEE
Transactions on Industrial Electronics

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS

Celsius in the WCS normal working conditions. So, the rectifier through the method proposed in Section III, they aim to achieve
equivalent load calculation results are shown in Fig.10, when the maximum efficiency and the rated output power at the same
the diode parameter values Vdio and Rdio change with the time, which is 3.3 kW for the developed EV wireless prototype.
junction temperature Tj. Fig.10 suggests that the change of the The experimental results have shown that system efficiency
junction temperature only has a small impact on the rectifier can reach as high as 93.5%, and output power is 3.33 kW,
equivalent load calculation results. Also, the previous analysis which is very close to the design target of the rated output
shows that the calculation method can achieve high accuracy, power 3.3 kW, on the condition of adopting the original
on conditions of different system load resistances, rectifier compensation capacitor values. However, when ignoring Le to
input inductances, DC voltages, and mutual-inductances. So, design the compensation capacitors, the experimental results
the proposed rectifier equivalent load calculation method has show that the measured system efficiency is 93.2%; while the
good robustness, in the cases of WCS parameter variations. measured output power has increased to 3.95 kW, which is
Besides, it should be noticed that in the practice case, the more than 600 W higher than the design target of the rated
equivalent resistance of the power battery will change in the output power. In this case, the reactive power in WCS will
charging process, and this will bring the variation of the increase, and bring large electric stresses to the system devices,
rectifier input impedance. Further work may be focused on the even affect the safe operation of WCS. Hence, these
dynamic process and features of both the WCS and the rectifier experiments have shown that the equivalent inductance of the
load during charging. rectifier load should be considered for system design and
analysis. Also, they have proved the effectiveness of the
C.Effects of Rectifier Load Equivalent Inductance
proposed compensation network design method, which can
Since the conventional design methods usually neglect the make WCS achieve high efficiency and rated output power.
equivalent inductance of the rectifier load [19-21], we are going
to discuss its effects. Fig.11 shows the simulation results of Le D.Verifications of Load Estimation Methods
effects, which are conducted by Matlab, according to the actual
parameter values of the developed EV wireless charging
prototype. These simulations have not considered the switching
losses of power converters, and other stray losses. So, the
values of the simulated output powers and efficiencies will be a
little larger than the experimental ones.

(a) Measured voltages used for secondary side load estimation.

Fig. 11. Simulation results of Le effects on output power and efficiency.

Fig.11 suggests the influence of Le on efficiency is very
small. However, the output power changes from about 3.4 kW
to more than 3.9 kW, when Le varies. Considering the rated
system output power 3.3 kW, the change range is more than
(b) Measured voltages used for primary side load estimation.
15%, under the influence of the equivalent inductance. Hence,
Fig. 12. Experimental results of the system voltages used for load
Le will significantly affect the output power, and the system estimations based on the standard parameter values.
output power will have obvious deviation with the rated one, if
the compensation networks are designed without considering Based on the developed EV wireless charging prototype, the
its influence. proposed secondary and primary side load estimation methods
In order to further prove this conclusion, we have designed can be verified. Fig.12 shows the measured system voltages
another group of compensation capacitors, based on the used for load estimations, under the condition of standard
conventional 8/π2 rectifier load calculation method and parameter values; where, measured voltages in Fig. 12(a) are
ignoring Le. Because the primary and secondary side used for secondary side load estimation, and the ones in Fig.
compensation network designs are decoupled, and Le only 12(b) are used for primary side load estimation. Through these
affects the design results in the secondary side as mentioned in experimental results, the positive zero crossing times can be
Section III, the secondary side compensation capacitors have directly extracted from the measured data, when the measured
been redesigned as follows: C2s=22.4 nF, C2p=73.3 nF; while voltage value turns from negative to positive. Also, the
the primary side compensation capacitors remain the same. fundamental amplitudes of the measured voltages can be
Since the original compensation capacitors are designed calculated by the FFT program in Matlab, to get the amplitude
of the fundamental voltage transfer function Gp. So according

0278-0046 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIE.2018.2793260, IEEE
Transactions on Industrial Electronics

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS

to the measured voltages, the amplitude and phase angle of Gp to 21.5 Ω, the phase angles of the voltage transfer functions
can be obtained, and the system load resistance can be become bigger and closer to 90°, so the measurement deviations
estimated through the process in the flow chart in Fig.13. will be more easily amplified by the tangent function in (21).
Moreover, the robustness of the proposed primary side load
estimation method is discussed towards parameter uncertainties.
The impact of the stray parameter is adopted to analyze the
robustness here. The major stray parameter in WCS is the
inductance of the connecting cables, which is not included in
the established system model. These cables are usually used to
connect coils to other parts, or be the lead wires of the inductors.
So, they mainly affect values of the coils and compensation
inductances. According to measurements of these cables, their
stray inductances are smaller than or about 1 uH. In order to
make its influences more obvious, we select this value as ±1.5
uH. Since the stray inductances are too small to be conducted in
experiments, simulations are adopted here. Taking the
conditions that L1 has the inductance with ±1.5uH errors as an
example, the simulation results are shown Fig.14. Further
calculation based on Fig.14 indicates that the estimation error
reaches to almost 6%, when L1 has the inductance with 1.5 uH
Fig. 13. Flow chart of the proposed load estimation method based on error. This error is obvious and the stray inductance should be
the measured primary side voltages. considered when estimating the system load. So, the connecting
TABLE II cables of the coils and the lead wires of the inductors are taken
LOAD ESTIMATION RESULTS USING PROPOSED SECONDARY AND PRIMARY into count, when we measured the parameter values. Since the
SIDE LOAD ESTIMATION METHODS
effects of the major stray parameter in WCS are considered, the
Prototype Secondary Side Load Primary Side Load proposed load estimation methods have good robustness, on
Parameter Values Estimation Method Estimation Method
conditions that stray parameters exist.
RL (Ω) Ls (uH) RL_Sesti (Ω) RL_Pesti (Ω)
83.3 41.3 43.5
42.9 113.9 41.5 44.9
49.1 39.4 44.6
83.3 21.2 24.4
21.5 113.9 21.3 24.7
49.1 20.7 24.1

Following the steps in Fig.13, the load estimation results can

be obtained and given in Tab.2, on conditions of different load
Fig. 14. Simulated load estimation results of the proposed primary side
resistances and rectifier input inductances. Tab.2 suggests that method, when L1 has the inductance with ±1.5uH errors.
for the secondary side load estimation method, the estimation
error is 3.7% on the condition of standard parameter values.
This means the proposed method has good accuracy, when the
system is well designed and working in the rated state. But
when Ls is changed to 49.1 uH, the estimation error has
increased to 8.2%, due to the DCM working state of the rectifier.
Also, when RL is changed to 21.5 Ω, the maximum estimation
error is only 3.7%. The estimation errors are mainly caused by
the measurement deviations of the voltage zero crossing times,
and the influences of the filter capacitor and its stray resistance, Fig. 15. Simulated load estimation results of the proposed primary side
which have been neglected in the initial value calculation method, when M varies about 23%.
process of the rectifier circuit. Moreover, for the primary side Furthermore, the robustness of the proposed primary side
load estimation method, Tab.2 suggests its estimation error is load estimation method is proved, when the mutual-inductance
1.4% on the condition of standard parameter values, which M is changeable. Through measurement and calculation, M is
means it also performs well. However, the estimation errors 19.5 uH with 15 cm lateral misalignment. In this case, M
obviously increase, on the conditions that RL is equal to 21.5 Ω. variation is more than 23%, when the lateral misalignment
This can be explained by the primary side load estimation distance changes from 0 cm to 15 cm. Based on the actual
method needs the measured parameter values of WCS, which parameter values, simulated load estimation results are shown
all have certain measurement deviations. When RL has changed in Fig.15. Further calculation indicates that the maximum

0278-0046 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIE.2018.2793260, IEEE
Transactions on Industrial Electronics

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS

estimation error is smaller than 3%, when M varies about 23%. has been established to calculate its equivalent input impedance,
Then, the experiments show that when the receive coil has a which contains both resistance and inductance components,
lateral misalignment distance of 10 cm, and other parameters and can be independently calculated through the parameters of
still have standard values, RL_Sesti is 41.7 Ω, and RL_Pesti is 44.7 Ω; the rectifier circuit. Based on the rectifier load analysis, a
while the estimation errors are 2.8% and 4.2%, respectively. compensation network design method is proposed to achieve
These experimental errors are also small. So, the proposed load the decoupling design of the primary and secondary side
estimation methods have good robustness, on conditions of compensation capacitors. Furthermore, a secondary side load
mutual-inductance M variation. estimation method and a primary side load estimation method
Finally, the robustness of the proposed primary side load are put forward, considering the influence of the rectifier load.
estimation method is analyzed, when the compensation They adopt only measured voltages to avoid the deviations
capacitances C1s, C1p, C2s, and C2p have 25% variations. Based introduced by different phase delays between measured voltage
on the actual parameter values, simulated estimation error and current. Finally, the established model, the proposed
percentages are shown in Fig.16, compared with the system rectifier load calculation method, compensation network design
load resistance 42.9 Ω. In Fig.16, the cases of compensation method, secondary and primary side load estimation methods
capacitances increasing by 25% are taken as examples. It is have been verified, based on the developed EV wireless
clear that the estimation errors are almost in the range of ±3%, charging prototype. The experimental results have shown the
when the compensation capacitances vary 25%. Then, the following conclusions: the equivalent input impedance of
experimental results are given, considering the related analysis rectifier load is mainly affected by system load resistance and
in Section V. C. When the secondary side compensation rectifier input inductance; rectifier load equivalent inductance
capacitances have been changed to 22.4 nF and 73.3 nF, and will impact system performances, and should be considered for
other WCS parameters still have standard values, RL_Sesti will be compensation network design; the proposed load estimation
40.9 Ω, and RL_Pesti will be 44.2 Ω; while the estimation errors methods have good accuracy, but still need to be improved in
are 4.7% and 3.0%, respectively. These experimental results further research; the proposed rectifier load calculation method
have shown that the maximum estimation error is less than 5%. and system load estimation methods all have good robustness,
So, the proposed load estimation methods have good robustness, on conditions of WCS parameter variations.
on conditions of compensation capacitance variations. Although the works in this paper are conducted based on the
specific system, they can be extended to more applications,
such as wireless charging systems with other rectifier or
compensation network topologies, etc. They will be helpful for
system design and control to make EV wireless charging
systems achieve stable operation and high performance.

REFERENCES
[1] S. Q. Li, and C. C. Mi, “Wireless power transfer for electric vehicle
applications,” IEEE J. Emerg. Sel. Topics Power Electron., vol. 3, no. 1,
Fig. 16. Simulated load estimation errors of the proposed primary side pp. 4-17, Mar. 2015.
method, when the compensation capacitances vary 25%. [2] Y. D. Ko, and Y. J. Jang, “The optimal system design of the online
electric vehicle utilizing wireless power transmission technology,” IEEE
Besides, when DC source voltage is changed to 200 V and Trans. Intell. Transp. Syst., vol. 14, no. 3, pp. 1255-1265, Sep. 2013.
other parameters still have standard values, RL_Sesti is 41.7 Ω, [3] W. X. Zhong, and S. Y. R. Hui, “Maximum energy efficiency tracking for
wireless power transfer systems,” IEEE Trans. Power Electron., vol. 30,
and RL_Pesti is 43.8 Ω; while the estimation errors are 2.8% and no. 7, pp. 4025-4034, Jul. 2015.
2.1%, respectively. So, these experimental results have shown [4] D. Ahn, S. Kim, J. Moon, and I. K. Cho, “Wireless power transfer with
that the proposed load estimation methods have good automatic feedback control of load resistance transformation,” IEEE
Trans. Power Electron., vol. 31, no. 11, pp. 7876-7886, Nov. 2016.
robustness, on conditions of DC source voltage variation. [5] M. P. Theodoridis, “Effective capacitive power transfer,” IEEE Trans.
To sum up, although there are some deviations, the estimated Power Electron., vol. 27, no. 12, pp. 4906-4913, Dec. 2012.
load resistances are basically close to the actual values of the [6] D. Thenathayalan, and J. H. Park, “Wide-air-gap transformer model for
the design-oriented analysis of contactless power converters,” IEEE
system load resistors. This has proved the effectiveness and Trans. Ind. Electron., vol. 62, no. 10, pp. 6345-6359, Oct. 2015.
robustness of the proposed secondary and primary side load [7] M. Fu, Z. Tang, M. Liu, C. Ma, and X. Zhu, “Full-bridge rectifier input
estimation methods. Further work may be focused on reducing reactance compensation in megahertz wireless power transfer systems,”
the measurement deviation effects of both the voltages and the in Proc. 2015 WoW, 2015, pp. 1-5.
[8] A. Berger, M. Agostinelli, S. Vesti, J. A. Oliver, J. A. Cobos, and M.
WCS parameters, in order to improve the accuracy of the load Huemer, “A wireless charging system applying phase-shift and amplitude
estimation methods, especially on conditions that phase angles control to maximize efficiency and extractable power,” IEEE Trans.
of the voltage transfer functions are close to 90°. Power Electron., vol. 30, no. 11, pp. 6338-6348, Nov. 2015.
[9] K. Colak, E. Asa, M. Bojarski, D. Czarkowski, and O. C. Onar, “A novel
phase-shift control of semibridgeless active rectifier for wireless power
VI. CONCLUSION transfer,” IEEE Trans. Power Electron., vol. 30, no. 11, pp. 6288-6297,
Nov. 2015.
This paper presents a systematic analysis of the rectifier load [10] Y. Akihara, T. Hirose, S. Masuda, N. Kuroki, M. Numa, and M.
used for EV wireless charging system. The rectifier load model Hashimoto, “Analytical study of rectifier circuit for wireless power
transfer systems,” in Proc. ISAP, 2016, pp. 338-339.

0278-0046 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/TIE.2018.2793260, IEEE
Transactions on Industrial Electronics

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS

[11] H. C. Li, K. P. Wang, L. Huang, W. J. Chen, and X. Yang, “Dynamic Yanjie Guo received the Ph.D degree in 2013
modeling based on coupled modes for wireless power transfer systems,” from the Institute of Electrical Engineering,
IEEE Trans. Power Electron., vol. 30, no. 11, pp. 6245-6253, Nov. 2015. Chinese Academy of Sciences (IEECAS).
[12] S. Aldhaher, P. C. K. Luk, K. E. K. Drissi, and J. F. Whidborne, Now he is an associate professor at the
“High-input-voltage high-frequency class E rectifiers for resonant Department of Vehicle Energy System and
inductive links,” IEEE Trans. Power Electron., vol. 30, no. 3, pp. Control Technology, IEECAS, and also at the
1328-1335, Mar. 2015. Key Laboratory of Power Electronics and
[13] H. T. Shi, J. D. Mao, X. S. Li, and J. T. Pan, “Modeling and analysis of Electric Drive, Chinese Academy of Sciences.
impedance for uncontrolled rectifier based nonlinear load,” in Proc. His research interests include wireless power
CCDC, 2016, pp. 1770-1775. transfer modeling and design, stationary
[14] C. K. Lee, S. Kiratipongvoot, and S. C. Tan, “High-frequency-fed unity wireless EV charging system, dynamic wireless
power-factor AC-DC power converter with one switching per cycle,” EV charging system, electromagnetic compatibility (EMC), power
IEEE Trans. Power Electron., vol. 30, no. 4, pp. 2148-2156, Apr. 2015. electronics applications, and electromagnetic field analysis.
[15] Q. Lei, M. Shen, V. Blasko, and F. Z. Peng, “A generalized input
impedance model of three phase diode rectifier,” in Proc. APEC, 2013, pp.
2655-2661. Lifang Wang (M’09) got her Ph.D degree in
[16] S. Q. Li, W. H. Li, J. J. Deng, T. D. Nguyen, and C. C. Mi, “A 1997 from Jilin University. After that she joined
double-sided LCC compensation network and its tuning method for the Institute of Electrical Engineering, Chinese
wireless power transfer,” IEEE Trans. Veh. Technol., vol. 64, no. 6, pp. Academy of Sciences (IEECAS). During the
2261-2273, Jun. 2015. Chinese tenth-five year plan (2001-2005), she
[17] M. Dionigi, M. Mongiardo, and R. Perfetti, “Rigorous network and was a member of the national specialist group
full-wave electromagnetic modeling of wireless power transfer links,” of Key Special Electric Vehicle Project of the
IEEE Trans. Microwave Theory Tech., vol. 63, no. 1, pp. 65-75, Jan. National 863 Program, and she was the head of
2015. the 863 Special EV Project Office.
[18] A. Costanzo, M. Dionigi, F. Mastri, M. Mongiardo, G. Monti, J. A. She is currently the director of Department
Russer, P. Russer, and L. Tarricone, “Conditions for a load-independent of Vehicle Energy System and Control
operating regime in resonant inductive WPT,” IEEE Trans. Microwave Technology at IEECAS. She is also the vice director of Key Laboratory
Theory Tech., vol. 65, no. 4, pp. 1066-1076, Apr. 2017. of Power Electronics and Electric Drive, Chinese Academy of Sciences.
[19] Q. W. Zhu, L. F. Wang, and C. L. Liao, “Compensate capacitor Her research interests include wireless charging system for EV, electric
optimization for kilowatt-level magnetically resonant wireless charging vehicle control system, EV battery management system,
system,” IEEE Trans. Ind. Electron., vol. 61, no. 12, pp. 6758-6768, Dec. electromagnetic compatibility, and smart electricity use. She has
2014. directed more than 15 projects in these fields and has published more
[20] J. W. Kim, D. H. Kim, and Y. J. Park, “Analysis of capacitive impedance than 90 papers and 30 patents.
matching networks for simultaneous wireless power transfer to multiple
devices,” IEEE Trans. Ind. Electron., vol. 62, no. 5, pp. 2807-2813, May
2015. Yuwang Zhang received the M.S. degree in
[21] J. Kim, and J. Jeong, “Range-adaptive wireless power transfer using electronic science and technology from the
multiloop and tunable matching techniques,” IEEE Trans. Ind. Electron., Wuhan University of Technology in 2015. He is
vol. 62, no. 10, pp. 6233-6241, Oct. 2015. currently working toward the Ph.D. degree at
[22] W. H. Li, H. Zhao, S. Q. Li, J. J. Deng, T. Z. Kan, and C. C. Mi, the Institute of Electrical Engineering, Chinese
“Integrated LCC compensation topology for wireless charger in electric Academy of Sciences (IEECAS).
and plug-in electric vehicles,” IEEE Trans. Ind. Electron., vol. 62, no. 7, His research interests include wireless
pp. 4215-4225, Jul. 2015. power transfer theory, optimization and control
[23] J. P. W. Chow, H. S. H. Chung, and C. S. Cheng, “Use of transmitter-side of electric vehicle wireless charging system,
electrical information to estimate mutual inductance and regulate bi-directional wireless charging system.
receiver-side power in wireless inductive link,” IEEE Trans. Power
Electron., vol. 31, no. 9, pp. 6079-6091, Sep. 2016.
[24] J. Yin, D. Lin, C. K. Lee, and S. Y. R. Hui, “A systematic approach for Shufan Li received his M.S. degree in electrical
load monitoring and power control in wireless power transfer systems engineering from the University of Chinese
without any direct output measurement,” IEEE Trans. Power Electron., Academy of Sciences in 2016. He is currently
vol. 30, no. 3, pp. 1657-1667, Mar. 2015. working toward the Ph.D. degree at the Institute
[25] Z. H. Wang, Y. P. Li, Y. Sun, C. S. Tang, and X. Lv, “Load detection of Electrical Engineering, Chinese Academy of
model of voltage-fed inductive power transfer system,” IEEE Trans. Sciences (IEECAS).
Power Electron., vol. 28, no. 11, pp. 5233-5243, Nov. 2013. His research interests include wireless
[26] M. F. Fu, H. Yin, X. Zhu, and C. B. Ma, “Analysis and tracking of optimal power transfer theory, parameter estimation and
load in wireless power transfer systems,” IEEE Trans. Power Electron., control of wireless charging system, dynamic
vol. 30, no. 7, pp. 3952-3963, Jul. 2015. wireless EV charging system.
[27] J. Yin, D. Lin, T. Parisini, and S. Y. R. Hui, “Front-end monitoring of the
mutual inductance and load resistance in a series-series compensated
wireless power transfer system,” IEEE Trans. Power Electron., vol. 31, Chenglin Liao (M’09) received the Ph.D.
no. 10, pp. 7339-7352, Oct. 2016. degree in Power Machinery and Engineering
[28] M. Feliziani, T. Campi, S. Cruciani, F. Maradei, U. Grasselli, M. from Beijing Institute of Technology, Beijing,
Macellari, and L. Schirone, “Robust LCC compensation in wireless China, in 2001. After that he had spent 2 years
power transfer with variable coupling factor due to coil misalignment,” in as a postdoctoral researcher at the Tsinghua
Proc. 2015 EEEIC, 2015, pp. 1181-1186. University.
[29] M. Liu, Y. Qiao, and C. B. Ma, “Robust optimization for a 6.78-MHz He is now the deputy director of Department
wireless power transfer system with class E rectifier,” in Proc. 2016 WoW, of Vehicle Energy System and Control
2016, pp. 88-94. Technology, Institute of Electrical Engineering,
[30] H. Feng, T. Cai, S. X. Duan, J. B. Zhao, X. M. Zhang, and C. S. Chen, Chinese Academy of Sciences (IEECAS). His
“An LCC-compensated resonant converter optimized for robust reaction currently research is mainly on the development
to large coupling variation in dynamic wireless power transfer,” IEEE of wireless charging system for electric vehicles.
Trans. Ind. Electron., vol. 63, no. 10, pp. 6591-6601, Oct. 2016.

0278-0046 (c) 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.