IEEE/CAA JOURNAL OF AUTOMATICA SINICA, VOL. 3, NO.

3, JULY 2016 281

Fractional Modeling and SOC Estimation of

Lithium-ion Battery
Yan Ma, Xiuwen Zhou, Bingsi Li, and Hong Chen, Senior Member, IEEE

Abstract—This paper proposes a state of charge (SOC) estima- these methods result in high errors caused by the accumulation
tor of Lithium-ion battery based on a fractional order impedance of errors in numerical integration in current measurement.
spectra model. Firstly, a battery fractional order impedance State observer[2−3] , Kalman filter (KF)[4−5] and particle filter
model is derived on the grounds of the characteristics of Warburg
element and constant phase element (CPE) over a wide range (PF)[6−7] are used to estimate SOC based on the model of
of frequency domain. Secondly, a frequency fitting method and Lithium-ion battery. The SOC estimation error of each method
parameter identification algorithm based on output error are is summarized in [8], which shows that the existing integer
presented to identify parameters of the fractional order model order battery SOC estimation methods mainly have estimation
of Lithium-ion battery. Finally, the fractional order Kalman filter error larger than 1 %, which may be because the models which
approach is introduced to estimate the SOC of the lithium-ion
battery based on the fractional order model. The simulation are obtained through the external characteristics of the power
results show that the fractional-order model can ensure an battery cannot show the precise internal characteristics. The
acceptable accuracy of the SOC estimation, and the error of dynamics of the battery is described by a set of integer order
estimation reaches maximally up to 0.5 % SOC. calculus equations. But complex electrochemical reactions are
Index Terms—Lithium-ion battery, fractional order model, described by the fractional order function.
electrochemical impedance spectra, fractional Kalman filter. The fractional order calculus (FOC) is a natural extension
of the classical integral order calculus. References [9−11]
I. I NTRODUCTION have shown that most phenomena, such as damping, friction,
mechanical vibration, dynamic backlash, sound diffusion, etc.,

G ENERALLY, the electrochemical reactions inside

lithium-ion battery are complicated in the running elec-
tric vehicle (EV), which is a highly nonlinear dynamic system.
have fractional order properties. Thus, FOC is widely used in
modeling, kinetics estimation, etc. FOC is also used to develop
the electrochemical models of the super capacitors and so on.
State of charge (SOC)[1] is defined as the percentage of the When it comes to FOC battery modeling and SOC estima-
amount of left energy to the rated capacity of a battery, which tion, [12] uses FOC model obtained by system identification
cannot be measured directly, it only can be estimated by to estimate crankability of battery, [13] proposes lead acid
measured variables such as current and terminal voltage. The battery state of charge estimation with FOC, and [14] deals
accurate estimation of SOC is the key problem in the field of with a fractional order state space model for the lithium-ion
power battery. battery and its time domain system identification method. The
The methods of SOC estimation are categorized into direct existing FOC modeling for battery meets the same problem,
experiment measurement methods and estimation methods the estimation accuracy in not high enough for battery man-
based on battery models. Coulomb counting method and cur- agement system.
rent integration method are the most popular experiment mea- The electrochemical impedance spectroscopy (EIS) method
surement methods, which are simple to obtain SOC. However, is one of the most accurate methods to model the electrochem-
Manuscript received August 31, 2015; accepted December 21, 2015. This ical Li-ion batteries. There are many studies which have tried
work was supported by National Natural Science Foundation of China (615 to utilize the impedance spectra directly to estimate SOC, but
20106008, U1564207, 61503149), High Technology Research and Develop- EIS method is too complicated to be used directly. EIS method
ment Program of Jilin (20130204021GX), Specialized Research Fund for
Graduate Course Identification System Program (Jilin University) of China is mainly used with equivalent circuit model at present[15−16] .
(450060523183), and Graduate Innovation Fund of Jilin University (20151 The remainder of the paper is organized as follows. Section
48). Recommended by Associate Editor YangQuan Chen. II discusses the battery fractional-order modeling based on
Citation: Yan Ma, Xiuwen Zhou, Bingsi Li, Hong Chen. Fractional mod-
eling and SOC estimation of Lithium-ion battery. IEEE/CAA Journal of
impedance spectra; Section III discusses how to obtain char-
Automatica Sinica, 2016, 3(3): 281−287 acteristic curve between open circuit voltage (OCV) and SOC,
Yan Ma is with the State Key Laboratory of Automotive Simulation and states order identification with frequency method and param-
Control, Jilin University, China, and also with the Department of Control
Science and Engineering, Jilin University (Campus Nanling), Changchun 1300
eters identification according to the output error identification
25, China (e-mail: [email protected]). algorithm; Section IV presents fractional order Kalman filter
Xiuwen Zhou and Bingsi Li are with Jilin University, Changchun 130012, for SOC estimation; Section V draws conclusions from the
China (e-mail: zhou [email protected]; [email protected]).
Hong Chen is with the State Key Laboratory of Automotive Simulation and
preceding work and offers suggestion for further study.
Control, Jilin University, Changchun 130012, China, and with the Department
of Control Science and Engineering, Jilin University (Campus Nanling), II. F RACTIONAL MODELLING OF BATTERY
Changchun 130025, China (e-mail: [email protected]). The impedance spectra curve of the Lithium-ion battery can
Color versions of one or more of the figures in this paper are available
online at http://ieeexplore.ieee.org. be got through Electrochemical workstation and is shown in
282 IEEE/CAA JOURNAL OF AUTOMATICA SINICA, VOL. 3, NO. 3, JULY 2016

 r
 d
Fig. 1. As shown in the figure, the impedance spectra can be  , r > 0,
 dtr
divided into three sections: the high frequency section, the mid r
∆ = 1, r = 0,
frequency section and the low frequency section. 

R
(dτ )r , r < 0.
The mathematical model in high frequency can be described
as (1).
V1 = R1 I, (1)
where R1 ∈ R denotes the value of Ohmic resistance, I
denotes the current, and V1 denotes the voltage of R1 .
The mathematical model in middle frequency can be de-
scribed as (2).
1 1
∆ β V2 = − V2 − I, (2)
R2 C2 C2
where C2 ∈ R is the coefficient of CPE, β ∈ R, −1 < β < 1
denotes the fractional order of CPE, R2 ∈ R denotes the value
of Ohmic resistance in ZARC element, I denotes the current
across the ZARC element, and V2 denotes the terminal voltage
of ZARC element.
Fig. 1. Impedance spectra of a Li-ion battery. The mathematical model of Warburg element in low fre-
quency can be described as (3).
In the high frequency section, the impedance spectra curve
1
intersects with the real axis and the intersection point could ∆α V3 = − I, (3)
be represented by an Ohmic resistance. W
In the low frequency section, the impedance spectra curve where W ∈ R is the coefficient, ∆α denotes α order of the
is a straight line with a constant slope, and has the same fractional element, α ∈ R, −1 < α < 1 is the fractional
impedance spectroscopy characteristic with constant phase order of Warburg element, V3 denotes the voltage of Warburg
element (CPE) which is usually referred to as a Warburg element, and I denotes the current across the Warburg element.
element. For the determined relationship of OCV and SOC, SOC
The middle frequency section forms a depressed semicircle, can be regarded as a system state, which can be presented as
which is a well-known phenomenon in electrochemistry. Such follows:
a depressed semicircle could be modeled by paralleling a 1
∆1 Soc = − I, (4)
Warburg element or CPE with a resistance, which is referred Qn
to as a ZARC element (it yields an arc in the Z plane)[17] . where Qn denotes the nominal capacity (Ah) of battery.
From the analysis above, the equivalent circuit model can be The relationship between SOC and OCV is nonlinear and
described as Fig. 2. Voc denotes open circuit voltage (OCV), it is not easy to draw a mathematical interpretation for it. It
and Vo denotes battery terminal voltage which can be directly is easy to find that when SOC is between 20 % and 80 % the
measured; R1 ∈ R denotes the value of Ohmic resistance, relationship is considered to be linear and can be written as
I denotes the current, and V1 denotes the voltage of R1 ; C2 follows:
∈ R is the coefficient of CPE in ZARC element, R2 ∈ R
denotes the value of Ohmic resistance in ZARC element, and Voc = k · Soc + d, (5)
V2 denotes the terminal voltage of ZARC element; W ∈ R is where k and d are the coefficients which can be calculated
the coefficient of Warburg element, and V3 denotes the voltage from the curve fitting.
of Warburg element. £ ¤T
Set the system state vector as x = V2 V3 Soc , the
system input as u = I, and the system output as y = Vo − d.
The continuous fractional state space function can be written
as (6).
(
∆N x = Ax + Bu,
(6)
y = Cx + Du,
   
− R21C2 0 0 − C12
 1 
where A =  0 0 0 , B =  − W , C =
Fig. 2. Fractional equivalent circuit model.
0 0 0 − Q1n
 
From the above description, the battery can be described by β
£ ¤
a fractional model. To simplify the FOC equation, we define 1 1 k , D = −R1 , N =  α .
the denotation as follows: 1
MA et al.: FRACTIONAL MODELING AND SOC ESTIMATION OF LITHIUM-ION BATTERY 283

According to the stochastic theory, the discrete state space A. Curve Fitting Between OCV and SOC
function is obtained as follows: OCV is obtained by fitting average value of charging and
(
∆N xk+1 = Axk + Buk , discharging terminal voltages which are measured by applying
(7) constant pulse current for each time 10 % SOC to battery in
yk = Cxk + Duk ,
both of charging and discharging modes.
where xk ∈ R3 denotes the system state vector, yk ∈ R Through the above test, the unknown parameters k and d
denotes the system output, and uk ∈ R denotes the system in (5) can be obtained by curve fitting.
input, all at the time instant k. The specific test procedure is as follows:
The fractional order Grünwald-Letnikov definition is given 1) Discharge the battery till it reaches the minimum dis-
as charging voltage (2 V in our case) at room temperature, and
k µ ¶ keep it idle for 12 hours.
N 1 X j N 2) Charge the battery with a constant current of 0.2 C (0.5 A)
∆ xk = N (−1) xk−j , (8)
Ts j till terminal voltage reaches 3.7 V. During the procedure, idle
j=0
the battery for 2 minutes after each 10 % SOC charging.
where Ts is the sample interval, and k is the number of Record every minimum voltage, as shown in Table I.
samples for which the derivative is calculated,
µ ¶ ( TABLE I
N 1, j = 0,
= N (N −1)···(N −j+1) MINIMUM POINTS OF EVERY SOC WHILE CHARGING
j , j > 0.
j! SOC (%) Voltage (mV)
Equation (9) can be derived from (8). 1 2664
µ ¶ 11 3140
N
xk+1 = TsN ∆N xk+1 − xk 21 3233
1 31 3278
k+1
X µ ¶
N 41 3304
+ (−1)j xk−j+1 . (9)
j 51 3318
j=2
61 3322
The discrete state space function of the battery can be 71 3341
written as (10). 81 3382
 91 3408

x N N
P
k+1
k+1 = Ts (A + N E)xk + Ts Buk − γj xk−j+1 , 100 3702
j=2

y = Cx + Du ,
k k k 3) Idle the battery for 12 hours.
(10) 4) Discharge the battery with a constant current of 0.2 C
½µ ¶µ ¶µ ¶¾ (0.5 A) till terminal voltage reaches 2 V. During the procedure,
β α 1
where γj = diag . idle the battery for 2 minutes after each 10 % SOC discharging.
j j j
Let Ad = TsN (A + N E), Bd = TsN B, Cd = C, Dd = D, Record every maximum voltage, as shown in Table II.
and E be unit matrix. Considering the process noise and output
TABLE II
noise, the discrete state space function of the system can be MAXIMUM POINTS OF EVERY SOC WHILE
written as DISCHARGING


x P
k+1
SOC (%) Voltage (mV)
k+1 = Ad xk + Bd uk + wk − γj xk−j+1 ,
j=2 (11) 99 3434

y = C x + D u + v ,
k d k d k k 89 3284
79 3274
where wk ∈ R3 is process noise, representing the modeling 69 3266
uncertainty and unknown input, vk ∈ R is output noise, on be- 59 3360
half of the measurement disturbance, wk and vk are assumed to 49 3240
be independent, zero mean Gaussian noise processes with the 39 3222
covariance matrices E[wk wjT ] = Qk δkj , E[vk vjT ] = Rk δkj , 29 3192
and δkj is Kronecker function. 19 3147
9 2812
III. PARAMETER IDENTIFICATION 0 2431
Parameter identification of battery model can be divided into
two sections, curve fitting of relationship between OCV and 5) Fit minimum points and maximum points that we col-
SOC and parameter identification in battery model. We will lected in prior experiments respectively and average the two
describe the two parts separately as following. curves, which are shown in Fig. 3.
284 IEEE/CAA JOURNAL OF AUTOMATICA SINICA, VOL. 3, NO. 3, JULY 2016

impedance spectra well, which means that the fractional-order

model can express the characteristic of Lithium-ion battery
well.
2) Unknown parameter identification: Unknown parame-
ters are identified via output error identification algorithm in
time domain[18−19] . The output error approach is diagrammed
in Fig. 6.

Fig. 3. OCV measurement during charging and discharging.

Fitting the OCV-SOC relationship from 20 % SOC to 80 %

SOC as shown in Fig. 4, we can get the values k = 0.002086
and d = 3.166.

Fig. 6. Parameter identification of battery by output error approach.

The transfer function of fractional order equivalent circuit
model shown in Fig. 1 can be written as
V (s) 1 R2 ( C21sβ )
H(s) = = · R 1 ·
I(s) W sα R2 + ( C21sβ )
Fig. 4. OCV-SOC curves between 20 % and 80 % SOC.
R 1 R2
= , (12)
W sα (1 + C2 R2 sβ )
B. Parameter Identification of Model
where V = Vo − Voc .
Identify unknown parameters in (6) with time domain and Equation (12) can be written as (13).
frequency domain method separately.
V (s)W sα (1 + C2 R2 sβ ) = I(s)R1 R2 . (13)
1) Identify the order of states using frequency fitting method
in frequency domain: This impedance spectra curve of the Applying of inverse Laplace transform algorithm, we have
Warburg element has the slope of απ/2. The slope of the dα dα+β
low-frequency part of the impedance spectra is nearly π/4, so W V (t) + C2 R2 W V (t) = R1 R2 I(t). (14)
dt dt
parameter α is equal to 0.5.  
The impedance spectra curve of the loop consisting of a
d
a CPE and a resistance is shaped like a semicircle. The Let a = R1 R2 , b = W , c = C2 R2 W , θ =  b , p = .
dt
regression rate of the semicircle will be changed with β. The c
Equation (14) is written as
bigger β is, the bigger the curve radian is. When β = 0.65,
the measured impedance spectra will be matched well, shown V (t) a B(θ)
G(p) = = α = . (15)
in Fig. 5. I(t) bp + cpα+β A(p, θ)
The noise-free output y(tk ) is supposed to be corrupted by
an additive white measurement noise v(tk ) which is normally
distributed with a zero mean and R variance, considered at
discrete instants. The complete equation can be written in the
form
(
y(tk ) = G(p) · u(tk ),
(16)
y ∗ (tk ) = y(tk ) + v(tk ),
where y ∗ (tk ) is the measured output of the system.
Assume that an error function ε(t) is given by the output
error, i.e.,
B(θ)
ε(t) = y ∗ (tk , θ) − u(t)
A(p, θ)
µ ∗ ¶ µ ¶
y (tk , θ) u(t)
= A(p, θ) − B(θ)
A(p, θ) A(p, θ)
Fig. 5. System state order fitting curve. = A(p, θ)yf∗ (t) − B(θ)uf (t), (17)
Fig. 5 shows that the impedance spectra curve obtained where yf∗ (t) = y ∗ (t, θ)/A(p, θ) and uf (t) = u(t)/A(p, θ).
by the order of state identification can match the measured Hence, a linear low-pass filter is applied to the measured
MA et al.: FRACTIONAL MODELING AND SOC ESTIMATION OF LITHIUM-ION BATTERY 285

input part and output part separately instead of a direct Bring the above parameters into (6), the discrete fractional
differentiation of the input variable and output variable. As model can be written as

shown in (17), the filter converts the output error into an  P
k+1
x
equation error[20] . k+1 = Ad xk + Bd uk + wk − γj xk−j+1 ,
j=2 (22)
Let F n (p) = 1/A(p, θn ), n = 1, 2, . . . stands for the 
 yk = Cd xk + Dd uk + vk ,
iteration number. In practical cases, A(p, θ) being unknown,
an estimation F n (p) = 1/Â(p, θ̂n ) is computed iteratively. where
   
The noise-free output variable is obtained through an aux- β − R21C2 0 0 −43289.39 0 0
iliary model, i.e., Ad =  0 α 0 =  0 0.5 0 ,
B̂(θ̂n ) 0 0 1 0 0 1
y n (t) = u(t). (18)  −1 
   
Â(p, θ̂n ) C2 −90.9 0.65
 −1 
The filtered input, output, and measured output are com- Bd =     
 W  = −0.038 , N = 0.5 ,

puted respectively with −1 −0.00011 1
£ Qn ¤ £ ¤
uf (t) = F n (p)u(t), Cd = 1 1 k = 1 1 0.21 , Dd = −R1 = −0.0243.
pα yf (t) = pα F n (p)y(t),
C. Model Validation
pα yf∗ (t) = pα F n (p)y ∗ (t).
The current profile consists of many charge/discharge
And they are gathered in the regression vectors as pulses, at different current levels. Economic Commission for
£ ¤T Europe (ECE) 15 urban driving cycle which is used on electric
ϕf (k) = uf (k) −pα yf∗ (k) −pα+β yf∗ (k) , vehicles is selected to simulate a typical driving pattern. The
£ ¤T
n α
ϕf (k) = uf (k) −p yf (k) −p α+β
yf (k) . current profile shown in Fig. 7 repeats the ECE 15 urban
driving cycle 3 times, and each circle is running for 400 s.
Thus, we have The voltage curve, shown in Fig. 8, includes the two curves.
£ ¤ One is the output of the identified model and the other is the
Φnf = ϕnf (0) ... ϕnf (Tfinal ) ,
£ ¤ measured voltage of the battery. And Fig. 9 shows the error of
Φf = ϕf (0) ... ϕf (Tfinal ) , the two voltage curves at different time. It is easy to find that
£ ¤
Yf∗ = yf∗ (0) ... yf∗ (Tfinal ) . (19) almost all the voltage errors are within 20 mV. When the input
The optimization problem of the parameter identification
can be stated as
° n ∗ °2
°
θ̂k = arg min °[Φnf ΦT
f ]θ − [Φf Yf ] . (20)
θ

The solution is given by

θ̂k = (Φnf ΦT
f)
−1 n ∗
Φf Yf , (21)
and the algorithm is iterated until convergence, when Fig. 7. Current profile for model validation.
max | θ̂k −θ̂θ̂k−1 | < ε , where ε is chosen by the accuracy of
k
modeling.
Specific identification process can be described as:
£ ¤T
1) k = 0, initialize the parameters with θ0 = 0 0 0 .
2) k = 1, 2, 3, . . ., calculate noise-free output y(k) accord-
ing to θk−1 and (16).
3) Filter the current, the terminal voltage, and the noise-free
terminal voltage based on (18).
4) Determine variables based on (19). Fig. 8. Voltage profile for model validation.
5) Update the identified parameters θ̂k by (21).
6) Calculate the relative error of | θ̂k −θ̂θ̂k−1 | from 2) to 5)
k
until the error is less than 0.05.
After the identification process, the value of every element
can be gotten as R2 = Ra1 , W = b, C2 = R2cW in (6), Qn
= Cn × 3600 where Cn is the nominal capacity, i.e., R2 =
2.1 mΩ, W = 26.5, C2 = 11 mF.
The intersection of impedance spectra with real axis in high
frequency shows the value of resistance. From the impedance
spectra curve we can get R1 = 24.3 mΩ. Fig. 9. Voltage error for model validation.
286 IEEE/CAA JOURNAL OF AUTOMATICA SINICA, VOL. 3, NO. 3, JULY 2016

charging current or discharging current switches largely, the Error covariance update
error is reaches to 40 mV.
Thus the identified fractional order model is accurate.
Pk = (1 − Kk C)P̃k . (27)
IV. SOC ESTIMATION BASED ON FRACTIONAL ORDER
MODEL 3) Save the estimated state and covariance for further
The work described in this paper was undertaken using iteration.
26 650 Lithium-ion batteries manufactured by A123 (2.5 Ah, 4) Separate the system state, and we will get SOC timely.
3.3 V batteries) which are shown in Fig. 10, and the battery The current profile shown in Fig. 12 is employed as vali-
test machine is shown in Fig. 11. dation scenario. Under this condition, fractional Kalman filter
and Kalman filter (KF) are used to estimate terminal voltage
and SOC of battery. Both the measured terminal voltage and
model terminal voltage are shown in Fig. 13. From Fig. 13, we
can see the battery is tested in full SOC range (i.e., terminal
voltage from 2.0 V to 3.6 V), and both the terminal voltage
Fig. 10. A123 cell.
estimated by FKF and KF can trace the measured terminal
voltage well, but the FKF results are more precise than the
KF ones.

Fig. 11. Battery test equipment.

Fig. 12. Current profile for SOC estimation.
A fractional order estimator is designed to estimate SOC of
the battery.
For the common integer order system, Kalman filter ap-
proach is widely used to estimate the parameters of the system.
Hence, fractional order Kalman filter (FKF)[21] is selected
to estimate the SOC of the battery.
The process can be described as:
1) k = 0
The Kalman filter is initialized with the best available
information of state and error covariance. The initialized value Fig. 13. Terminal voltage profile for SOC estimation.
of state estimation and error covariance are expressed as:
the covariance matrices of process noise Q, the covariance SOC estimation and SOC estimation error curves are shown
matrices of measurement noise R, the initialized system state in Fig. 14 and Fig. 15, respectively. From Fig. 15 we will see,
x̂0 , and the covariance matrices of initialized system state P0 at the beginning of battery charging, the FKF estimation error
= E[(x̂0 − x0 )(x̂0 − x0 )T ]. is almost the same as KF estimation error. And with further
2) k = 1, 2, . . . charging of battery, precision of FKF will rise gradually while
State estimation propagation precision of KF gets worse. At the late charging period, both
X k µ ¶ the FKF and KF error are increasing, but the FKF estimation
N
x̃k = Ad xk−1 + Bd Ik − γj xk−j . (23) error is always smaller than the KF one. In the whole test, the
j
j=1 error of FKF can be reduced up to max 0.5 % SOC, while the
Error covariance propagation error of KF reaches 3 %.
k
X
P̃k = (Ad + N1 )Pk−1 + Qk−1 + Nj Pk−j NjT . (24)
j=2

Kalman gain update

Kk = P̃k C T (C P̃k C T + Rk ). (25)
State estimation update
x̂k = x̃k + Kk (yk − C x̃k ). (26) Fig. 14. SOC estimation.
MA et al.: FRACTIONAL MODELING AND SOC ESTIMATION OF LITHIUM-ION BATTERY 287

[15] Luo Y F, Gong C S A, Chang L X, Liu Y H. AC impedance technique

for dynamic and static state of charge analysis for Li-ion battery. In:
Proceedings of the 17th IEEE International Symposium on Consumer
Electronics (ISCE). Hsinchu, China: IEEE, 2013: 9−10
[16] Eddahech A, Briat O, Bertrand N, Delétage J Y, Vinassa J M. Behavior
and state-of-health monitoring of Li-ion batteries using impedance
spectroscopy and recurrent neural networks. International Journal of
Electrical Power and Energy Systems, 2012, 42(1): 487−494
[17] Barsoukov E, Macdonald J R. Impedance Spectroscopy: Theory, Experi-
Fig. 15. SOC estimation error. ment, and Applications (2nd edition). Hoboken, NJ: Wiley-Interscience,
2005.
V. C ONCLUSION [18] Guermah S, Djennoune S, Bettayeb M. Discrete-time fractional-order
systems: modeling and stability issues. Advances in Discrete Time
Based on the analysis of the impedance spectra, a simplified Systems. INTECH Open Access Publisher, 2012.
battery fractional-order model is derived. A new identification [19] Guermah S, Djennoune S, Bettayeb M. Controllability and observability
method is presented to identify the orders of the states and of linear discrete-time fractional-order systems. International Journal of
Applied Mathematics and Computer Science, 2008, 18(2): 213−222
parameters based on the fractional-order system. The fractional [20] Victor S, Malti R, Garnier H, Oustaloup A. Parameter and differentiation
Kalman filter is utilized to estimate the SOC of the lithium-ion order estimation in fractional models. Automatica, 2013, 49(4): 926−
battery based on the fractional-order model. The simulation 935
results show the SOC estimation with fractional Kalman filter [21] Sierociuk D, Dzieliński A. Fractional Kalman filter algorithm for the
states, parameters and order of fractional system estimation. Interna-
is consistent with expectations. However just one battery is tional Journal of Applied Mathematics and Computer Science, 2006,
tested in the paper, battery pack will be tested in further study. 16(1): 129−140

R EFERENCES Yan Ma received the B. S. degree in the Automatica

[1] Wang Y, Zhang C, Chen Z. A method for joint estimation of state- Department from Harbin Engineering University,
of-charge and available energy of LiFePO 4 batteries. Applied Energy, China, in 1992, M. S. and Ph. D. degrees in the De-
2014, 135: 81−87 partment of Control Science and Engineering from
[2] Kim D, Koo K, Jeong J J, Goh T, Sang W K. Second-order discrete- Jilin University, China, in 1995 and 2006. In 1995,
time sliding mode observer for state of charge determination based on a she joined former Jilin University of Technology.
dynamic resistance Li-ion battery model. Energies, 2013, 6(10): 5538− She has been a post doctor at Poly University, Hong
5551 Kong. Since 2009, she has been a professor at Jilin
University. Her current research interests include
[3] Chen X, Shen W X, Cao Z, Kapoor A. Sliding mode observer for state of
nonlinear estimation methods and applications in
charge estimation based on battery equivalent circuit in electric vehicles.
power management system of EV, and robust filter
Australian Journal of Electrical and Electronics Engineering, 2012, 9(3):
methods.
225−234
[4] Li D, Ouyang J, Li H, Wan J F. State of charge estimation for
LiMn 2O 4 power battery based on strong tracking sigma point Kalman Xiuwen Zhou graduated from Jilin University,
filter. Journal of Power Sources, 2015, 279: 439−449 China, in 2013. She is currently a master student
in the Department of Control Science and Engineer-
[5] He Ling-Na, Wang Yun-Hong. SOC estimation based on improved sam-
ing, Jilin University, China. Her research interests
pling point Kalman filter for mine-used battery. Journal of Mechanical
include battery management system of EV.
and Electrical Engineering, 2014, 31(9): 1213−1217 (in Chinese)
[6] Wang Y J, Zhang C B, Chen Z H. A method for state-of-charge
estimation of LiFePO 4 batteries at dynamic currents and temperatures
using particle filter. Journal of Power Sources, 2015, 279: 306−311
[7] Li J H, Barillas J K, Guenther C, Danzer M A. Multicell state estimation
using variation based sequential Monte Carlo filter for automotive
battery packs. Journal of Power Sources, 2015, 277: 95−103
Bingsi Li graduated from Jilin University, China,
[8] Lu L G, Han X B, Li J Q, Hua J F, Ouyang M G. A review on the key
in 2014. She is currently a master student in the
issues for Lithium-ion battery management in electric vehicles. Journal
Department of Control Science and Engineering,
of Power Sources, 2013, 226: 272−288
Jilin University, China. Her research interests in-
[9] Deng Li-Wei, Song Shen-Min. Synchronization of fractional order clude battery management system of EV.
hyperchaotic systems based on output feedback sliding mode control.
Acta Automatica Sinica, 2014, 40(11): 2420−2427 (in Chinese)
[10] Sadli I, Urbain M, Hinaje M, Martin J P, Raël S, Davat B. Contributions
of fractional differentiation to the modelling of electric double layer
capacitance. Energy Conversion and Management, 2010, 51(12): 2993−
2999
[11] Fairweather A J, Foster M P, Stone D A. Battery parameter identification
Hong Chen received the B. S. and M. S. degrees
with pseudo random binary sequence excitation (PRBS). Journal of
in process control from Zhejiang University, China,
Power Sources, 2011, 196(22): 9398−9406
in 1983 and 1986, respectively, and Ph. D. de-
[12] Sabatier J, Cugnet M, Laruelle S, Grugeon S, Sahut B, Oustaloup A, gree from the University of Stuttgart, Germany,
Tarascon J M. A fractional order model for lead-acid battery cranka- in 1997. In 1986, she joined Jilin University of
bility estimation. Communications in Nonlinear Science and Numerical Technology, China. From 1993 to 1997, she was a
Simulation, 2010, 15(5): 1308−1317 “Wissenschaftlicher Mitarbeiter” at the Institut Fuer
[13] Sabatier J, Aoun M, Oustaloup A, Grégoire G, Ragot F, Roy P. Fractional Systemdynamik und Regelungstechnik, University
system identification for lead acid battery state of charge estimation. of Stuttgart. Since 1999, she has been a professor
Signal Processing, 2006, 86(10): 2645−2657 at Jilin University, where she serves currently as
[14] Wu H J, Yuan S F, Yin C L. A Lithium-Ion battery fractional order state Tang Aoqing Professor. Her current research inter-
space model and its time domain system identification. In: Proceedings ests include model predictive control, optimal and robust control, nonlinear
of the 2012 FISITA World Automotive Congress. Berlin Heidelberg: control and applications in process engineering and mechatronic systems.
Springer, 2013, 192: 795−805 Corresponding author of this paper.