See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/366278961

Constrained integral fuzzy sliding mode control for an electric vehicle: An

LMI-based approach

Article in Transactions of the Institute of Measurement and Control · December 2022

DOI: 10.1177/01423312221138888

CITATIONS READS

0 25

4 authors:

Saeed Amiri Seyed Mohsen Seyed Moosavi

National Iranian South Oil Company Islamic Azad University of Ahvaz
3 PUBLICATIONS 1 CITATION 7 PUBLICATIONS 67 CITATIONS

SEE PROFILE SEE PROFILE

Mehdi Forouzanfar Ebrahim Aghajari

Università della Calabria Islamic Azad University of Ahvaz
32 PUBLICATIONS 312 CITATIONS 21 PUBLICATIONS 122 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Saeed Amiri on 23 January 2023.

The user has requested enhancement of the downloaded file.

Article – Measurement
Transactions of the Institute of
Measurement and Control
Constrained integral fuzzy sliding 1–12
Ó The Author(s) 2022

mode control for an electric vehicle: Article reuse guidelines:

sagepub.com/journals-permissions
DOI: 10.1177/01423312221138888
An LMI-based approach journals.sagepub.com/home/tim

Saeed Amiri , Seyed Mohsen Seyed Moosavi,

Mehdi Forouzanfar and Ebrahim Aghajari

Abstract
This study investigates the issue of stability analysis and controller synthesis for an electric vehicle based on the Takagi–Sugeno fuzzy control approach.
The proposed adaptive fuzzy integral sliding mode controller ensures that the electric vehicle system performs satisfactorily in the presence of uncer-
tainties and exogenous disturbances, comprising low computational load, small control input amplitude, and reduced energy consumption. For efficient
performance of the electric vehicle, the battery voltage (control input) and the vehicle’s speed (system output) are considered restricted. Simulations
of an electric vehicle powered by a brushed direct current motor are utilized to evaluate the efficacy of the proposed controller. Ultimately, the design
procedure validates the robust performance and fast stabilization of the electric vehicle speed in the presence of exogenous disturbances and para-
metric uncertainties.

Keywords
Electric vehicle, adaptive fuzzy integral sliding mode controller, Takagi–Sugeno fuzzy model, adaptive mechanism, input–output constraints

Introduction the sector nonlinearity (SN) method. The TS model is a uni-

versal approximator because it can approximate any smooth
In light of the finite nature of fossil energy sources and the nonlinear function. In addition, the TS model is a convex
rapid pace at which this energy is used for transportation, the combination of linear models that expedites stability analysis,
vehicle sector is compelled to seek alternative energy sources. complex system modeling, controller design, and other related
Electric vehicles (EVs) have gained well-deserved popularity concepts within the context of linear matrix inequalities
worldwide, and the global EV industry has made a tremen- (LMIs) (Ghalehnoie et al., 2020; Lendek et al., 2011; Sun
dous stride forward in the previous decade (Hmidi et al., et al., 2019). Fuzzy logic–based control of EV subsystems has
2020; Huin et al., 2021; Zhang et al., 2020). The layout of the come out as one of the most strong and prolific aspects of
substantial EV’s powertrain and control system is shown in inquiry in recent years (Mohammadzadeh et al., 2021).
Figure 1. A controller is one of the most important compo- Sliding mode control (SMC), a subclass of variable struc-
nents of an EV, as it is responsible for directing the vehicle ture control systems, is one of the most extensively used non-
and producing a high-performance EV with an ideal highest linear control systems (Utkin, 1977). SMC methods are
speed balance, accelerated performance, and traveling range known for their strong robustness and insensitivity to para-
per charge. The control system is the top priority in EV safety metric uncertainty and exogenous disturbance. Integral slid-
and plays a vital role in an electric motor lifetime, minimizing ing mode control (ISMC) methods in combination with TS
the maintenance costs, increasing the battery efficiency, and systems have been frequently employed in recent years
ensuring optimal driving performance. Direct current (DC) (Echreshavi et al., 2021a; Hwang and Kim, 2020; Yu and
motors have some significant advantages; these motors have Weng, 2014). To achieve high performance and robustness
largely been used for speed and torque control due to their throughout the closed-loop system and accelerate the reaching
simplicity. The brushed direct current (BDC) motor is one of phase, the idea of ISMC was proposed. The major
the DC motors that is used broadly in EVs. Generally speak-
ing, road topology conditions and motor parameters are typi-
cally changing, and hence the presence of exogenous Department of Electrical Engineering, Ahvaz Branch, Islamic Azad
University, Iran
disturbance is inevitable (Khooban et al., 2016; Singh et al.,
2020). Corresponding author:
During the previous decades, the Takagi–Sugeno (TS) Seyed Mohsen Seyed Moosavi, Department of Electrical Engineering,
fuzzy model has attracted significant attention because it is a Ahvaz Branch, Islamic Azad University, Farhangshahr, Golestan Blvd,
potent tool for modeling, stability, and control of nonlinear Ahvaz 61349-37333, Iran.
systems with complexity. TS model of a system is achieved by Email: m.moosavi@srbiau.ac.ir
2 Transactions of the Institute of Measurement and Control 00(0)

Figure 1. EV powertrain and control configuration.

disadvantage of using SMC is two related phenomena: chat- performance and reduce energy consumption. The problem of
tering and excessive control action activity, both of which can restrictions on the battery voltage (control input) and the
contribute to a reduction in control performance or even speed of EV (system output) in the framework of LMIs is con-
instability of the closed-loop control system. It has been sidered as the impetus of this study. Even though the ISMC
demonstrated that TS fuzzy control and higher-order sliding issue for the TS fuzzy systems of the EV has been investigated
modes, particularly the ISMC, complement each other a little and some promising solutions have been published
because fuzzy control provides a strong framework for expert (Goggia et al., 2014; Sharma and Singh, 2019), there are a few
knowledge representation. In contrast, ISMC techniques have problems to consider: (1) some previous studies did not
some valuable advantages, which are robust against exogen- include practical problems such as LMI-based limitations on
ous disturbances, model uncertainties, and parameter varia- the battery voltage and the EV’s speed, which are essential
tions (Pan et al., 2017; Zhang et al., 2021). A new fuzzy power considerations; (2) for simulating the topological road condi-
management algorithm is based on a super-twist control strat- tions and motor settings, the disturbance signal applied to the
egy used on a battery-super-capacitor system for EV, compris- EV with unknown structure is not taken into account; and (3)
ing power supervision algorithm and control of the traction the constrained adaptive fuzzy integral sliding mode (AFISM)
portion (Sellali et al., 2019). The H‘ lateral controller, controller is used to overcome the simultaneous uncertainty
according to Guo et al. (2020), is intended to mitigate the and exogenous disturbances. This method yields additional
impact of exogenous disturbance and uncertainty in electric design challenges when the systems employ the TS fuzzy
four-wheel vehicles by using a fuzzy observer to estimate the model.
roll effects and actuator faults. The aim of this paper is to discuss the subject of con-
In Khatun et al. (2003), through refinement of the fuzzy strained TS-ISMC for a class of continuous-time uncertain
membership functions, a fuzzy controller is suggested for EV nonlinear systems that are exposed to uncertainties and
wheel slip control in an EV’s antilock braking system. A com- exogenous disturbances. The following are the most signifi-
bined SMC algorithm controls the regenerative braking force cant contributions made by this paper:
of the electric motor to avoid the wheels from locking up
when the motor is braking (He et al., 2020). In order to
(1) To ensure efficient control, both the battery voltage
achieve sufficient energy savings and brake effectiveness dur-
of the EV, used as the control input, and the EV
ing braking circumstances, the controller is developed using
speed, used as the system output, are restricted.
the sliding-mode fuzzy control technique (Mei et al., 2021). In
(2) Construct an integral fuzzy sliding surface for the EV
Zhang et al. (2016), a constant voltage charging controller
system.
has been designed to provide high performance and robust
(3) Ensure uniformly ultimate boundedness of the
regenerative charging by combining the SMC technique with
closed-loop system of the EV.
the TS approach. In order to reduce the impacts of exogenous
disturbance torque and model uncertainty in DC motors, an
optimal controller is developed (Ruderman et al., 2008). A The remainder of this work is structured in the following
fuzzy H‘ controller is intended to overcome the actuator fail- manner. In the ‘‘EV modeling and preliminaries’’ section, the
ures, control input restrictions, and sprung variation of in- problem formulation and preliminaries will be presented in
wheel motor-driven EVs (Shao et al., 2017). four sections. The TS model is presented in this part of the
It seems that less attention has been given to the issue of paper. The new integral sliding surface and AFISM controller
input–output limitations on the EVs in order to improve are presented in the ‘‘Main results’’ section. In the section
Amiri et al. 3

where TL denotes the torque of the EV, re is the tire’s radius,

and G is motor-to-axle gear ratio (Huang et al., 2010;
Larminie and Lowry, 2012).

EV motor model
The electric motor is a vital part of an EV. These motors are
widely applied in EVs. The BDC motors have an easy struc-
ture of electric motors, with broad usage. It is typical to model
a BDC motor as (Al Halabi and Al Tarabsheh, 2020)
  
Figure 2. The forces applied on a EV moving up a slope. ðL1 + L2 Þ didtE = V  Rma + Rmf iE  L12 iE wE
ð3Þ
J dw 2
dt = L12 iE  BwE  T
E

‘‘Specification of input–output performance,’’ the limited

specifications of the input and output of the EV are demon- In equation (3), iE denotes the armature current, and wE is
strated. The simulation results and conclusion have been given the motor’s angular speed. The inertia of the motor is denoted
in the sections ‘‘Simulation results’’ and ‘‘Conclusion.’’ by the J . Rma , Rmf , L1 , and L2 are considered as the armature
resistance, field winding resistance, armature inductance, and
field winding inductance, respectively, T is exogenous torque,
EV modeling and preliminaries B is the viscous coefficient, L12 is the mutual inductance of the
armature and field windings, and the control input is denoted
EVs are a less complex structure than internal combustion
by V . For the sake of simplicity, x = ½ iE wE T = ½ x1 x2 T .
engine (ICE) vehicles. The initial phase toward modeling an EV
Afterward, equation (4) depicts the relationship between the
is studying the dynamics of an EV and the motor dynamics.
motor’s angular speed w and the driving velocity n
Electric motors drive EVs via transmission and differential gears,
which ultimately provide power to the wheels. In this section, the re
tractive effort, motor model, overall dynamic model, and eventu- n= x2 ð4Þ
G
ally the TS model of the EV are represented.

Model of tractive force Entire dynamic model of the EV

In this section, with combining the two formulas in equations
As shown in Figure 2, the modeling of an EV incorporates
(1) and (3), the total dynamic model of the EV system with
the balance of forces occurring on a moving vehicle. These
series BDC motor can be written as
include the aerodynamic drag force F1 , acceleration of the
vehicle F2 , rolling drag force F3 , and hill-climbing forces F4 .   " ðRm a + Rm f Þ L12
#   1 
The combined effect leads to a tractive effect force Fte . The d x1 L1 + L2 L1 + L2 x1 x1
= L12 + L1 + L2 V ð5Þ
aerodynamic drag force is the initial exogenous force on the dt x2 J + mr2 =G2 :x1 P x2 0
EV. This force is owing to the friction of the air when the EV
is traveling. The EV’s linear acceleration will be provided by where
the acceleration force. The rolling drag force is determined by
the amount of friction that exists between the tires of the EV 1
P= re 2
and the road surface. This force is equal to the EV’s mass. J +m
G
 r 3
The force required to drive the EV up a slope is the most e 1 r 
e
r 
e
straightforward to find, is hill-climbing force. The entire trac- B+0:5rACd x2 + mrr mg +mg sin (a)
G x2 G G
tive effort is equal to the sum of all forces, as previously stated
and as specified in equation (1).
The parameter values for the EV are given in Table 1 (Sharma
1 and Purwar, 2014).
FT (t) = F1 + F2 + F3 + F4 = mrr mg + rACd n2
2 ð1Þ
dn
+ mg sin a + m TS model of the EV
dt
Consider the continuous-time TS model as:
where mrr is the rolling drag factor, m denotes the EV’s mass, Fuzzy Rule i: IF ,1 (t) is Mi1 (t),..., and ,p (t) is Mip (t), THEN
r is the air density, g is the gravity, A is the frontal, Cd is the

drag factor, a is the hill-climbing angle, and n is the driving x_ (t) = Ai x(t) + Bi u(t)
velocity of the EV. The relation between the entire force and ð6Þ
y(t) = Ci x(t), i = 1, 2, :::, r
the torque is expressed by equation (2)
Here, Mij are membership functions ( j= 1, 2, ...,p, p is the
Fte re number of the premise variables (PVs)). x(t) 2 Rn , u(t) 2 Rm ,
TL = ð2Þ
G and y(t) 2 Rn denote the state vector, input vector, and output
4 Transactions of the Institute of Measurement and Control 00(0)

Table 1. The parameters of the EV system. approach, determine the minimal and maximal values of ,1 (t)
and ,1 (t) under ,1 (t) 2 ½d1 , D1  and ,2 (t) 2 ½d2 , D2  as follows
Parameters Values Parameters Values
8 8
>
> minf,1 (t)g = 0 >
> maxf,1 (t)g = 250
Rma + Rmf (O) 0.2 mrr 0.015 < <
x1 (t), x2 (t) x1 (t), x2 (t)
L1 + L2 (mH) 6.008 a 0 ð10Þ
>
> minf, 2 (t)g = 0:0065 >
> maxf, 2 (t)g = 267:3
L12 (mH) 1.776 r (kg=m3 ) 1.25 : :
x1 (t), x2 (t) x1 (t), x2 (t)
J (kg:m2 ) 0.05 A (m2 ) 1.8
m (kg) 800 Cd 0.3
By using the minimal and maximal values in equation (10),
re (m) 0.25 V (Volt) 0~48
G 11 i (A) 78
,1 (t) and ,2 (t) can be calculated by
B (N:M:s) 0.0002 wnorm (r=min) 2800 
,1 (t) = M1 (,1 (t)) maxf,1 (t)g + M2 (,1 (t)) minf,1 (t)g
,2 (t) = M3 (,2 (t)) maxf,2 (t)g + M4 (,2 (t)) minf,2 (t)g
vector, repectively. Ai 2 Rn 3 n , Bi 2 Rn 3 m , and Ci 2 Rq 3 m are  ð11Þ
M1 (,1 (t)) + M2 (,1 (t)) = 1
state matrix, input matrix, and output matrix, respectively, r
M3 (,2 (t)) + M4 (,2 (t)) = 1
being the number of the model rules. Moreover, ,1 (t),...,,p (t) are
PVs of the fuzzy rules, and input matrix Bi has full column rank, By employing equations (10) and (11), the membership
that is, rank(Bi ) = m\n, that the pairs (Ai , Bi ) are controllable. functions can be determined in the following manner
Based on equation (6), the total fuzzy model is as follows
8
8 >
> M1 (,1 (t)) = 0:004,1 (t), Large
> P
r <
>
< x_ (t) = li (,(t))(Ai x(t) + Bi u(t)) M2 (,1 (t)) = 0:004(250  ,1 (t)), Smaller
ð12Þ
i=1
ð7Þ >
> M (, (t)) = 0:004(,2 (t)  0:0065), Larger
> P
r : 3 2
>
: y(t) = li (,(t))Ci x(t), i = 1, 2, :::, r M4 (,2 (t)) = 0:004(267:2727  ,2 (t)), Small
i=1

Q Consequently, the fuzzy rules are expressed as follows

where ,(t) = ½,1 (t) ,2 (t) ::: ,p (t)T , ci (,(t)) = pj= 1 Mij (,j (t)), Rule R1 : if ,1 (t) is Large and ,2 (t) is Larger, Then
li (,(t)) = Prci (,(t)) . li (,(t)) is denoted as the normalized x_ (t) = A1 x(t) + B1 u(t),
i=1
ci (,(t)) Rule R2 : if ,1 (t) is Large and ,2 (t) is Small, Then
weight of the if–then rules. The term Mij (,j (t)) are member- x_ (t) = A2 x(t) + B2 u(t),
ship functions of fuzzy sets. Rule R3 : if ,1 (t) is Smaller and ,2 (t) is Larger, Then
x_ (t) = A3 x(t) + B3 u(t),
Remark 1. The normalized membership functions li (,(t)) Rule R4 : if ,1 (t) is Smaller and ,2 (t) is Small, Then
indicate the degree to which the ith-related local model is x_ (t) = A4 x(t) + B4 u(t),
activated. This function indicates how much the local related Here, the matrices A1 , A2 , A3 , and A4 are achieved as follows
model contributes to the overall model. It establishes a steady
2 3 2 3
transition between the current model and its neighboring ðRma + Rmf Þ L12 d1 ðRma + Rmf Þ L12 d1
A1 = 4
L1 + L2 L1 + L2 5A2 = 4 L1 + L2 L1 + L2 5
local models. These functions, which are in general dependent L12 d1 ðB + d2 Þ L12 d1 ðB + D2 Þ
on the state vector x(t), are triangular, sigmoidal, or J + mðrGe Þ
2
J + mðre 2
G Þ J + mðrGe Þ
2
J + mð re 2
GÞ
Gaussian in shape and exhibit the following convex charac-
teristics (Echreshavi et al., 2021b) 2 3 2 3
ðRma + Rmf Þ L12 D1 ðRma + Rmf Þ L12 D1
L +L L1 + L2 5A4 = 4 L1 + L2 L1 + L2
A3 = 4 L1 12 D1 2 ðB + d2 Þ L12 D1 ðB + D2 Þ
5
8 r J + mðrGe Þ
2
J + mðre 2
Þ J + mðrGe Þ
2
J + mð re 2
Þ
<P G G

li (,(t)) = 1
ð8Þ 23
: i=1
li (,(t)) ø 0, i = 1, 2, :::, r 1
B1 = B2 = B3 = B4 = 4 5
L1 + L2
To form the TS model of the EV system, the SN approach 0
has been presented. This is one of the most frequently used
techniques for fuzzy control design, since it may provide an The fuzzy system with applied disturbance to the EV is speci-
accurate fuzzy representation of complicated systems. fied as
Considering this feature, the SN approach can be applied to 8
> P
r
EV systems. Then, applying the SN approach and defining >
< x_ (t) = li (,(t))(Ai x(t) + Bi u(t) + Ei q(t))
the following PVs to equation (5) yields i=1
P
r ð13Þ
>
>
( : y(t) = li (,(t))Ci x(t), i = 1, 2, :::, r
,1 = n
iE i=1
 3  r e   o
,2 = 0:5rACd rGe wE + 1
wE mrr mg G + mg sin (a) rGe In equation (13), Ei is a column vector of disturbance, and
ð9Þ q(t) depicts the exogenous disturbance signal with the uncer-
tain structure, which is unknown but bounded and satisfied as
Due to the fact that the system states are bounded, the PVs
equation (9) will be bounded as well. Then, based on the SN k q(t) k ł ∂1 + ∂2 k x(t) k ð14Þ
Amiri et al. 5

where ∂1 and ∂2 are unknown parameters derived from the  and matrix
exists a symmetric positive definite (PD) matrix P
q(t) and x(t) structure. The intention of this study is to present  which gives
K,
a new SMC method based on the EV’s TS dynamic system
 + Bi K
 Þ + ðAi P
 + Bi K
 Þ \0, T
(13) that takes into account the uncertain and disturbance ðAi P (i = 1, 2, :::, r) ð19Þ
components. Furthermore, simple control techniques and
conditions with less design conservatism adapt to a vast vari- The matrix K then becomes
ety of control systems.
P
K =K  1 ð20Þ
Remark 2. For the AFISM controller design of the EV sys-
tem (13), we consider two cases with different matrices: the Proof. Consider the Lyapunov candidate V = xT (t)Px(t) for
Ei = Bi case and Ei 6¼ Bi . Accordingly, in subsequent sections, equation (13), where PT = P 2 R(n 3 n) and P  = P1 .
we will investigate the stability and reachability criteria for Subsequently, the time derivative of V is computed as
sliding-motion dynamics using a fuzzy sliding surface equa-
tion (15) in two distinct situations. X
r
V_ (t) = li (,(t))xT (Ai + Bi K)T Px(t)
i=1

Main results X
r
+ li (,(t))xT (t)(Ai + Bi K)T x(t)
This section will discuss the fuzzy surface and the AFISM i=1

controller design process for the EV system (13). The X

r
= li (,(t))xT (t)(Ai + Bi K)T x(t)
Lyapunov stability technique is applied to evaluate the reach- i=1
ability and stability criteria. X
r
+ li (,(t))xT (t)(Ai + Bi K)T x(t)
i=1
Fuzzy sliding surface X
r
=  Ai T + Ai P
li (,(t))xT (t)fP  +P
 + Bi K P  KPi gx(t)
The following fuzzy surface is offered to facilitate the control- i=1
ler design stage via the use of an integral type sliding surface Xr
(Fridman et al., 2005) =  i T + Ai P
li (,(t))xT (t) PA  +P
 + Bi K P  K T Bi T x(t)
i=1
ðt X
r Xr
=  + Bi K)
li (,(t))xT (t)f(Ai P  + (Ai P  T gx(t)
 + Bi K)
s(x, t) = =x(t)  li (,(u))=(Ai + Bi K)x(u)d(u) ð15Þ
i=1
i=1
0 Xr
=  + Bi K)
li (,(t))xT (t)f(Ai P  + (Ai P  T gx(t)
 + Bi K)
where the matrix = 2 Rm 3 n is non-singular, and matrix i=1
K 2 Rm 3 n is chosen so that the matrix Ai + Bi K being ð21Þ
Hurwitz. It should be noted that compared to the traditional
SMC, ISMC impressively improves transient performance in According to equation (18), it concludes V_ \0. There exists
an uncertain disturbed environment. The aim of ISMC is to an adequately small scalar g satisfying
compel the sliding motion on the surface s(x, t) = 0, thus by n o
derivative of surface equation (15) leads to  + Bi K
ðAi P  Þ + ðAi P  ÞT + g\0
 + Bi K ð22Þ
X
r
s_ (x, t) = li (,(t))=Bi fu(t) + q(t)  Kx(t)g ð16Þ Thus
i=1
V_ (x(t)) ł  g k Px(t)k2 ł  glmin (P) k x(t)k2 ð23Þ
For both s = s_ = 0, the equivalent control can be stated as
where lmin (P), is the lowest eigenvalue of matrix P.
ueq (t) = Kx(t)  q(t) ð17Þ Consequently, the quadratic stability of equation (19) can be
ensured via Lemma 1 in equation (18).
The following sliding motion can be achieved by substituting
equation (17) into equation (13)
AFISM controller design for disturb EV system
X
r
x_ (t) = li (,(t))ðAi + Bi K Þx(t) ð18Þ In this section, the AFISM controller aims to drive the state
i=1 trajectory of the system equation (13) into the sliding surface
equation (15), and afterward to maintain them within the
Since the matrix Ai + Bi K is stable in the sense of Hurwitz
sliding surface despite uncertainties or disturbances. In fact,
stability for each i, the sliding motion can be ensured to be
the AFISM controller adjusts the sliding variables. In other
stable in conformity with the following Lemma 1. The gain K
words, setting conditions on the controller for the reachability
selected from feasible solutions in Lemma 1.
of the sliding surface makes the surface attractive. For the
Lemma 1. (Teixeira and Zak, 1999). The sliding-motion purpose of designing an AFISM controller against uncertain-
dynamic in equation (18) is quadratically stabilizabled if there ties and disturbances, it is necessary to have the upper bands
6 Transactions of the Institute of Measurement and Control 00(0)

of them. Although, it is not easy to obtain these upper significant. Inspired by the above, in this part, we consider
bounds. Thus, the adaptive mechanism is utilized to estimate designing the AFISM controller for an uncertain model of
known parameters ∂^1 (t) and ∂^2 (t) to the unknown system the EV system with exogenous disturbance.
parameters ∂1 and ∂2 , respectively
Remark 3. The parameter uncertainties addressed here are
∂~1 (t) = ∂^1 (t)  ∂1 ð24Þ norm-bounded because, DAi (t) = Ti Di (t)Hi , Ti and Hi are real
known constant matrices with proper dimensions, and Fi (t) is
∂~2 (t) = ∂^2 (t)  ∂2 ð25Þ
an obscure time-varying matrices with Lebesgue-measurable
elements and satisfying DTi (t)Di (t) ł I, in which I is an unit
Theorem 1. The subsequent law equation (26) is intended to matrix (Zhu et al., 2020).
ensure that the closed-loop system of the EV equation (13),
integral manifold, and all signals are stable in the sense of Based on the above descriptions and formula (13), the
uniformly ultimate bounded (UUB)
P (Lee and Zak, 2004). fuzzy system for the EV is described in the following
With defining the Oi (t) = ri = 1 li (,(t))=Bi yields
X
r
s(t)Oi (t)  ^  x_ (t) = li (,(t))f(Ai + DAi )x(t) + Bi u(t) + Ei q(t)g ð31Þ
u(t) = ∂1 (t) + ∂^2 (t) k x(t) k + Kx(t)
k s(t)Oi (t) k ð26Þ i=1

 ufsign(s(t)Oi (t))  eg
The fuzzy surface is selected as
The adaptive laws are designed as
ðt X
r
  s = =x(t)  li (,(u))=ðAi + DAi + Bi K Þx(u)d(u) ð32Þ
∂^_ 1 (t) = r1 k sT (t)Oi (t) k d1 ∂^1 (t) ð27Þ i=1
0
 
∂^_ 2 (t) = ð1  r2 Þ k sT (t)Oi (t) kk x(t) k d2 ∂^2 (t) ð28Þ Based on equation (32) and the TS model of the EV equa-
tion (31), the derivative of fuzzy ISMC is gained as
where r1 , r2 , d1 , d2 , u are designed parameters and ∂^1 (0) . 0
and ∂^2 (0) . 0.
X
r
s_ = li (,(t))=Bi fu(t)  Kx(t)g ð33Þ
Proof. Define the Lyapunov function candidate as follows i=1

when the state trajectory of equation (31) enters the sliding

1 1 ~2 1 2 motion, the equivalent control is achieved as
V (t) = sT (t)s(t) + ∂1 (t) + ∂~2 (t) ð29Þ
2 2r1 2(1  r2 )
ueq (t) = Kx(t) ð34Þ
Taking the time derivative of V based on equations (16) and
(26), it results

1 1 ~ ^_
V_ (t) = sT fOi (t)ðu(t) + q(t)  Kx(t)Þg + ∂~1 (t)∂^_ 1 (t) ∂2 (t)∂2 (t)
r1 1  r2
   
= sT (t)Oi (t)fu(t) + q(t)Kx(t)g + ∂~1 (t) k sT (t)fOi (t) k d1 ∂^1 (t) + ∂~2 (t) sT (t)Oi (t) kk x(t) k d2 ∂^2 (t)

s(t)Oi (t)  ~ 
ł k sT (t)Oi (t) k  ∂1 (t) + ∂~2 (t)x(t)  u(sign(s(t)Oi (t))  e) + ∂1 + ∂2 k x(t) k
k s(t)Oi (t) k
 
+ ∂~1 (t) + ∂~2 (t) k x(t) k k sT (t)Oi (t) k
2 2
1 1 1
=  u k sT (t)Oi (t)k1 + eu k s(t)Oi (t) k d1 ∂^1 (t)  ∂1  d2 ∂^2 (t)  ∂2 + d1 ∂21 + d2 ∂22
2 2 4
1 1
ł  u k s(t)Oi (t)k1 + eu k s(t)Oi (t)k1 + d1 ∂21 + d2 ∂22 ð30Þ
4 4

Since k s(t)Oi (t)k1 ø k s(t)Oi (t) k , when k s(t)Oi (t) k6¼ 0

and based on equation (30) with using Theorem 1, the UUB
of system (13) can be obtained.
then, by substituting equation (34) into equation (31) and
AFISM controller design for uncertain EV system applying =Ei = 0 in Remark 2, the following sliding-motion
dynamic is obtained
Model uncertainties and disturbances always exist in practical
applications, since information about the system is generally X
r
insufficient. Consequently, considering the parameter uncer- x_ (t) = li (,(t))fðAi + Bi K Þx(t) + Ei q(t)g ð35Þ
tainties and disturbances in the EV system scheme is i=1
Amiri et al. 7

In accordance with the sliding motion defined by equation V_ (t) = sT (t)fOi (t)(u(t) + q(t)  Kx(t))g
(35), the subsequent Definition 1 and Theorem 2 present the 1 1
+ ∂~1 (t)∂^_ 1 (t) + ∂~2 (t)∂^_ 2 (t)
bounded stability criterion of equation (35). r1 2(1  r2 )
= sT (t)Oi (t)fu(t) + q(t)  Kx(t)g + ∂~1 (t)
Remark 4. The matrix K can be selected for the proposed  T 
fuzzy surface equation (15) in such a manner that Ai + Bi K is k s (t)Oi (t) k d1 ∂^1 (t) + ∂~2 (t)
 T 
Hurwitz for two different cases: Bi = Ei and Bi 6¼ Ei . In addi- k s (t)Oi (t) kk x(t) k d2 ∂^2 (t)
tion, the matrix K can be chosen from a feasible solution of ł k sT (t)Oi (t) k
(
conditions equations (19) and (36), respectively. s(t)Oi (t)  ^ 
 ∂1 (t) + ∂^2 (t) k x(t) k
Definition 1. Consider the system (35). Let k s(t)Oi (t) k
D )
x(t) 2 f ¼ fj :k j k ł d1 , j 2 Rn g for bounded parameter
d1 ø 0, d1 2 R, and t ø 0. If limx!‘ x(t) = 0 holds, for all usign(s(t)Oi (t)) + ∂1 (t) + ∂2 (t) k x(t) k
x(t) 2 j, therefore equation (35) is said to be boundedly    
stable. + ∂~1 (t)+∂~2 (t) k x(t) k k sT (t)Oi (t) k d1 ∂^1 (t) ∂^1 (t)∂1
 
 d2 ∂^2 (t) ∂^2 (t)  ∂2
Theorem 2. System (35) is stated as boundedly stable for 2
given PD symmetric matrix Qi if there exist a feasible solution 1
=  u k s(t)Oi (t)k1  d1 ∂^1 (t)  ∂1
(P, K) with PD matrix P = PT 2 Rn 3 n and matrix K m 3 n such 2
2
that satisfying 1 1  2 
 d2 ∂^2 (t)  ∂2 + d1 ∂1 + d2 ∂22
2 4
 1 
lmin ðQi Þ  2∂2 k PEi k . 0
D ð36Þ ł  u k s(t)Oi (t)k1 + d1 ∂21 + d2 ∂22
Qi ¼ ðAi + Bi K ÞT P  PðAi + Bi K Þ 4
ð39Þ
Proof. With using the Lyapunov function candidate If k s(t)Oi (t) k ł e, according to the control law deter-
V (t) = xT (t)Px(t) for the EV system mined in equation (38) and the adaptive mechanism specified
in equations (39) and (40) results

X
r X
r  2
V_ (t) = T T
li (,(t))fAi +Bi K)Ei q(t)g Px(t)+x (t)P li (,(t)) V_ (t) = sT (t)Oi (t) (s(t)Oi (t)) ∂^1 (t) + ∂^2 (t) k x(t) k
i=1 i=1
fAi + Bi K g + Ei q(t)g usign(s(t)Oi (t)) + ∂1 + ∂2 k x(t) k
X
r ð40Þ
= li (,(t))fxT (t)(Ai + Bi K)T P + P(Ai + Bi K)x(t) + ∂~1 (t)( k sT (t)Oi (t) k d1 ∂^1 (t)) + ∂~2 (t)
 T 
i=1 k s (t)Oi (t) kk x(t) k d2 ∂^2 (t)
+ qT (t)Px(t) + xT (t)PEi qT (t)g 1 
ł  u k s(t)Oi (t)k1 + d1 ∂21 + d2 ∂22 + 1
X
r 4
ł li (,(t))f2(∂1 + ∂2 k x(t) k )PEi k ) k x(t) k
i=1 Consequently, on the basis of equations (39) and (40), and
 lmin (Qi ) k x(t)k2 g based on Lee and Zak (2004), the UUB of the EV system (31)
X
r can be obtained.
= li (,(t))f(lmin (Qi )  2∂2 k PEi k ) k x(t)k2
i=1
+ 2∂1 k PEi kk x(t) kg Specification of input–output performance
ð37Þ It is possible that the permissible control approach will equip
an EV with sufficient energy management capabilities. For
Based on equation (36), V_ (t)\0, therefore equation (35) is this purpose, the mathematical framework based on LMI con-
boundedly stable. In accordance with equation (30), the straints on EV’s speed and the battery voltage is presented.
AFISM controller for the unertain EV system is designed as The critical characteristic of the proposed controller is to pro-
follows vide less voltage domain and activity as the control input and
8 quick EV speed stabilization with the minimum dissipated
s(t)Oi (t)
 
>
>  ks(t)O ∂^1 (t) + ∂^2 (t) k x(t) k + Kx(t) energy as the system output to achieve an efficient control.
>
< i (t)k
usign(s(t)Oi (t)) if k s(t)Oi (t) k . e Assume the initial condition is specified, then the EV’s restric-
u(t) =  2 ð38Þ tions k V k2 ł C and k wk2 ł Y as applied to the battery and
>
> s(t)O ^ ^
i (t) ∂1 (t) + ∂2 (t) k x(t) k + Kx(t)
>
: speed, respectively, are enforced at all times t ø 0 if the follow-
usign(s(t)Oi (t)) if k s(t)Oi (t) k ł e
ing LMIs hold (Farinwata et al., 2001)
If k s(t)Oi (t) k . e, according to the control law deter-     " #
1 x(0)T Q WiT Q QCiT
mined in equation (38), the adaptive mechanism defined in ø 0, ø 0, ø 0 ð41Þ
2
equations (24) and (25) results x(0) Q Wi C I Ci Q Y2 I
8 Transactions of the Institute of Measurement and Control 00(0)

Algorithm 1 (Fuzzy SMC steps)

Step 1. Define nonlinear terms (5) in terms of ,1 (t) and ,2 (t) with using formulas in (9).
Step 2. Determine the maximum and minimum values of the nonlinear terms (9).
Step 3. The terms M1, M2, M3, and M4 can be taken as membership functions (12) and thus, determine the normalized firing strengths l1 , l2 , l3 ,
and l4 .
Step 4. Define the fuzzy rules.
Step 5. Determine the state matrices A1 , A2 , A3 , and A4 based on the maximum and minimum values of the ,1 (t) and ,2 (t) and fuzzy rules.
Step 6. For the TS model in (13) two cases with different matrices (with Ei = Bi or Ei 6¼ Bi ) are considered.
Step 7. Design the SMC law in (26) or (30) as the control input (battery).
Step 8. Determine the sliding surface matrix K in two cases from a feasible solution of (18) for case Ei = Bi and a feasible solution of inequality (34)
for case Ei 6¼ Bi which represent in Example 1 and 2.
Step 9. Choose design parameters e, r1 , r2 , u, d1 , d2 , ∂^1 , and ∂^2 presented in controllers and adaptive laws.
Step 10. Applying the designed control input on the EV system (12) with control input (battery) and system output (EV’s speed).
Step 11. Calculation of energy consumption.

where Wi = Ki Q and Q = P1 , Ci 2 R1 3 2 . before, the output matrix Ci = ½0 1, and the parameters’ cal-
culated values C = 48 and Y = 2700. Then, matrix K is
Remark 5. (Vazquez-Sanchez et al., 2012). In actuality, BDC selected as K = ½114:72  215:14 by utilizing a feasible solu-
motors are constantly subject to disturbances and various  P)
tion (K,  and equation (8) with considering Lemma 1, the
sorts of noise, such as those caused by the BDC motor itself, fuzzy surface can be obtained as follows
those connected to the power source, and those produced by
adjacent components. Disturbances may cause the following: ð4 X
r
(a) false or double pulses in a BDC motor’s current signal s = =xðtÞ  li (,(u))=ðAi + Bi K Þx(u)du ð43Þ
and (b) ghost or merging pulses in the BDC motor current. i=1
0

The stages below are outlined to provide a quick overview All the simulations are implemented by MATLAB soft-
of the suggested process: ware and the SeDuMi solver. The initial condition is
x(0) = ½ 70 1000 T and the exogenous disturbance
q(t) = 1 + sin (t). The gain K is obtained as
EV energy consumption K = ½114:7162  215:1433 and thus, Figure 3(a) and (b)
describe the armature current and angular speed of motor
Implementing principles related to energy management is nec- responses of the EV system, respectively. The fuzzy surface
essary in order to achieve the goals of reducing energy con- and control signal responses are shown in Figure 3(c) and (d),
sumption and emissions while enhancing EV’s overall respectively. After applying new LMI conditions on the
efficiency. Numerous elements, such as EV speed, ambient input–output system (41), the matrix K is obtained as
temperature, control method, battery charge state, EV mass, K = 1:0e  03½0:0378  0:1525. As a result, Figure 4(a) and
and so on have a significant impact on the EV’s energy usage. (b) depicts the armature current and angular speed of the EV
The average amount of energy consumed is system’s motor responses, respectively. The fuzzy surface and
ð ð control signal responses are shown in Figure 4(c) and (d),
Eave = Ec Dl = Pc dt Vl dt ð42Þ respectively. Hence, the designed controller reduces external
disturbances while being robust to structural uncertainties,
has a small amplitude control input, and low values of x1
where Ec denotes the overall consumption (wh), the EV’s
(Armature current) in restricted case in Figure 4 lead to low
travel distance is Dl (km), and Pc and Vl stand for the EV’s
energy consumption features. As compared to Figures 3 and
power and speed, respectively. The findings of the simulation,
4 in constrained cases, the control action amplitude and chat-
which are depicted in Table 2, can be applied to calculate the
tering are reduced, and the control performance is improved.
average amount of energy consumed.
On the contrary, the controller’s design is restricted to achiev-
ing rapid EV speed stabilization while using the least amount
Simulation results of energy.
Example 2. In this example, based on system (31), the effect
This section includes two simulation examples to illustrate the
of uncertainty will be removed by utilizing a suitable surface.
suggested methods’ efficacy and superiority.
For the designed AFISM controller in equation (38), we
Example 1. Applying the designed AFISM controller in equa- select u = 11:33, r1 = 0:02, r2 = 0:01, d1 = 0:04, d2 = 0:02,
tion (26) for the TS fuzzy of EV in equation (12), we select e = 0:01, E1 = E3 = ½0 0:2T , E2 = E4 = ½0 0:1T , and matrix
u = 5:5, r1 = 0:01, r2 = 0:01, d1 = 0:04, d2 = 0:03, e = 1, = = ½0 1 for case Bi 6¼ Ei .
and matrix = = 0.1 0.2 for case Bi = Ei . The adaptive para- The adaptive parameters are selected as ∂^1 (0) = 0:1 and
meters are given as ∂^1 (0) = 0:2 and ∂^2 (0) = 0:3. As mentioned ∂^2 (0) = 0:2. The matrix K is selected as
Amiri et al. 9

Table 2. Comparison of energy usage. manifold and control input reveal in Figure 6(a) and (b). By
 P)
obtaining a solution (K,  and Lemma 1, the fuzzy surface
Case Unrestricted Restricted Unrestricted Restricted can be obtained
Example 1 3.23 wh 2.44 wh 1.92 wh 1.77 wh ðt X
4
Example 2 3.88 wh 2.67 wh 2.22 wh 1.94 wh
s = =x(t)  li (,(u))=ðAi + DAi + Bi K Þx(u)d(u) ð44Þ
i=1
0

K = 1e  3½0:0380  0:1469 and hence, Figure 5(a) and (b) Assume that the initial condition is x(0) = ½ 70 1200 T
depict the armature current and angular speed of motor and q(t) = 1 + 0:5 sin (t) is exogenous disturbance. The uncer-
responses of the EV system, respectively. After introducing tain matrices in Remark 3 are chosen as
new LMI limitations to the system’s input–output configura-
tion, the gain is K = 1e  3½0:0380  0:1469 and the fuzzy D1 = D2 = D3 = D4 = sin (t)

Figure 3. Stabilization of unrestricted responses: (a) armature current (ampere), (b) angular speed of motor (r/s), (c) control input, and (d) fuzzy
sliding surface.

Figure 4. Stabilization of restricted responses: (a) armature current (ampere), (b) angular speed of motor (r/s), (c) control input, and (d) fuzzy
sliding surface.
10 Transactions of the Institute of Measurement and Control 00(0)

Figure 5. Stabilization of unrestricted uncertain responses: (a) armature current (ampere), (b) angular speed of motor (r/s), (c) control input, and
(d) fuzzy sliding surface.

Figure 6. Stabilization of restricted uncertain responses: (a) armature current (ampere), (b) angular speed of motor (r/s), (c) control input, and
(d) fuzzy sliding surface.

   
0:1 0 0:1 0 The restricted control input in Figure 6(a) results in a rela-
T1 = , T2 = ,
0:1 0 0 0:1 tively limited amplitude of the control input (battery voltage).
    In comparison to Figures 5 and 6, the amplitude of the con-
0:1 0 0 0
T3 = , T4 = trol input and chattering is decreased, whereas the control
0:1 0 0 0:1
performance is enhanced. Due to the LMI restriction on the
    battery voltage amplitude and a considerable reduction in
0:01 0 0:01 0
H1 = , H2 = , states, energy consumption will be decreased. On the con-
0:01 0 0 0:01 trary, the designed controller is restricted to providing fast
   
0:01 0 0:01 0 EV speed stabilization with the least amount of energy con-
H3 = , H4 =
0:01 0 0:01 0 sumed. The simulation results shown in Table 2 can be used
Amiri et al. 11

to calculate the amount of energy consumption. This com- magnitude and rate constraints. Journal of the Franklin Institute
parison is shown clearly in Table 2, which demonstrates the 358(1): 575–605.
reduced energy utilization for restricted cases in Examples 1 Farinwata SS, Filev D and Langari R (2001) Fuzzy control: Synthesis
and 2. and analysis. Computing and Control Engineering Journal 12(2):
95–95.
Fridman L, Poznyak A and Bejarano F (2005) Decomposition of the
min–max multi-model problem via integral sliding mode. Interna-
Conclusion tional Journal of Robust and Nonlinear Control 15(13): 559–574.
Ghalehnoie M, Akbarzadeh-Tootoonchi MR and Pariz N (2020)
In this paper, the TS approach for modeling the EV system Fuzzy control design for nonlinear impulsive switched systems
has been investigated. Utilizing this approach, the nonlinear using a nonlinear Takagi-Sugeno fuzzy model. Transactions of the
components of the system can be replaced by linear subsys- Institute of Measurement and Control 42(9): 1700–1711.
tems, and the results are expressed as a simplified model with Goggia T, Sorniotti A, De Novellis L, et al. (2014) Torque-vectoring
straightforward dynamics. By selecting the Lyapunov func- control in fully electric vehicles via integral sliding modes. In: 2014
tion, a suitable fuzzy sliding surface, and designed AFISM American control conference, Portland, OR, 4–6 June.
controller, the EV system is stable in the sense of UUB so that Guo J, Wang J, Luo Y, et al. (2020) Robust lateral control of autono-
the controllers, armature current, the angular speed of the mous four-wheel independent drive electric vehicles considering
the roll effects and actuator faults. Mechanical Systems and Signal
motor, and states are stable with defining constraints on the
Processing 143: 106773.
battery voltage (control input) and EV’s speed (system out-
He L, Ye W, He Z, et al. (2020) A combining sliding mode control
put). In addition, the suggested controller ensures the EV sys- approach for electric motor anti-lock braking system of battery
tem’s satisfactory performance, including low computational electric vehicle. Control Engineering Practice 102: 104520.
load, minimum control input amplitude, and low energy con- Hmidi ME, Ben Salem I and El Amraoui L (2020) An efficient
sumption in the presence of uncertainties and exogenous dis- method for energy management optimization control: Minimizing
turbances. Sufficient conditions and constraints to ensure the fuel consumption for hybrid vehicle applications. Transactions of
efficient control of the EV system in terms of LMIs are pre- the Institute of Measurement and Control 42(1): 69–80.
sented. As a consequence, the benefits and advantages of the Huang Q, Li J and Chen Y (2010) Control of electric vehicle. Urban
proposed methodology have been proven via simulation Transport and Hybrid Vehicles. Epub ahead of print 18 August
2010. DOI: 10.5772/10184.
results. Recently, several novel model predictive control find-
Huin X, Di Loreto M, Bideaux E, et al. (2021) Improving perfor-
ings (Khooban et al., 2017; Vafamand et al., 2019) for EV mance specifications of internal combustion engine dedicated to
nonlinear systems have been suggested. We will refer to these plug-in hybrid electric vehicles based on coupled optimization
findings in future work and continue to study fuzzy SMC methodology. Transactions of the Institute of Measurement and
model predictive control challenges for EV nonlinear systems. Control. Epub ahead of print 25 July. DOI: 10.1177/
01423312211029692.
Hwang S and Kim HS (2020) Extended disturbance observer-based
Declaration of conflicting interests integral sliding mode control for nonlinear system via T–S fuzzy
The author(s) declared no potential conflicts of interest with model. IEEE Access 8: 116090–116105.
respect to the research, authorship, and/or publication of this Khatun P, Bingham CM, Schofield N, et al. (2003) Application of
article. fuzzy control algorithms for electric vehicle antilock braking/trac-
tion control systems. IEEE Transactions on Vehicular Technology
52(5): 1356–1364.
Funding Khooban MH, Niknam T and Sha-Sadeghi M (2016) Speed control
of electrical vehicles: A time-varying proportional–integral
The author(s) received no financial support for the research, controller-based type-2 fuzzy logic. IET Science, Measurement &
authorship, and/or publication of this article. Technology 10(3): 185–192.
Khooban MH, Vafamand N, Niknam T, et al. (2017) Model predictive
control based on Takagi-Sugeno fuzzy model for electrical vehicles
ORCID iDs delayed model. IET Electric Power Applications 11(5): 918–934.
Saeed Amiri https://orcid.org/0000-0003-0234-0028 Larminie J and Lowry J (2012) Electric Vehicle Technology Explained.
Mehdi Forouzanfar https://orcid.org/0000-0002-6440- Hoboken, NJ: John Wiley & Sons.
5731 Lee Y and Zak SH (2004) Uniformly ultimately bounded fuzzy adap-
tive tracking controllers for uncertain systems. IEEE Transactions
on Fuzzy Systems 12(6): 797–811.
References Lendek Z, Guerra TM, Babuska R, et al. (2011) Stability Analysis and
Nonlinear Observer Design Using Takagi-Sugeno Fuzzy Models.
Al Halabi M and Al Tarabsheh A (2020) Modelling of electric vehi-
Berlin; Heidelberg: Springer.
cles using MATLAB/SIMULINK. SAE technical paper no. 2020-
Mei P, Yang S, Xu B, et al. (2021) A fuzzy sliding-mode control for
01-5086. Available at: https://doi.org/10.4271/2020-01-5086
regenerative braking system of electric vehicle. In: 2021 7th inter-
Echreshavi Z, Farbood M and Shasadeghi M (2021a) Fuzzy event-
national conference on control, automation and robotics (ICCAR),
triggered integral sliding mode control of nonlinear continuous-
Singapore, 23–26 April.
time systems. IEEE Transactions on Fuzzy Systems 30: 2347–2359.
Mohammadzadeh A, Ahmadian A, Elkamel A, et al. (2021) Chaos
Echreshavi Z, Shasadeghi M and Asemani MH (2021b) HN dynamic
synchronization of brushes direct current motors for electric
observer-based fuzzy integral sliding mode control with input
 12 Transactions of the Institute of Measurement and Control 00(0)

vehicle: Adaptive fuzzy immersion and invariance approach. Teixeira MC and Zak SH (1999) Stabilizing controller design for
Transactions of the Institute of Measurement and Control 43(1): uncertain nonlinear systems using fuzzy models. IEEE Transac-
178–193. tions on Fuzzy Systems 7(2): 133–142.
Pan Y, Yang C, Pan L, et al. (2017) Integral sliding mode control: Utkin V (1977) Variable structure systems with sliding modes. IEEE
Performance, modification, and improvement. IEEE Transactions Transactions on Automatic Control 22(2): 212–222.
on Industrial Informatics 14(7): 3087–3096. Vafamand N, Arefi MM, Khooban MH, et al. (2019) Nonlinear
Ruderman M, Krettek J, Hoffmann F, et al. (2008) Optimal state-space model predictive speed control of electric vehicles represented by
control of DC motor. IFAC Proceedings Volumes 41(2): 5796–5801. linear parameter varying models with bias terms. IEEE Journal of
Sellali M, Abdeddaim S, Betka A, et al. (2019) Fuzzy-super twisting Emerging and Selected Topics in Power Electronics 7(3): 2081–2089.
control implementation of battery/super capacitor for electric Vazquez-Sanchez E, Gomez-Gil J, Gamazo-Real JC, et al. (2012) A
vehicles. ISA Transactions 95: 243–253. new method for sensorless estimation of the speed and position in
Shao X, Naghdy F and Du H (2017) Reliable fuzzy HN control for active brushed DC motors using support vector machines. IEEE Trans-
suspension of in-wheel motor driven electric vehicles with dynamic damp- actions on Industrial Electronics 59(3): 1397–1408.
ing. Mechanical Systems and Signal Processing 87: 365–383. Yu W-S and Weng C-C (2014) H tracking adaptive fuzzy integral slid-
Sharma U and Singh B (2019) A generalised double integral slid- ing mode control for parallel manipulators. Fuzzy Sets and Sys-
ing mode control for bidirectional charger of light electric vehi- tems 248: 1–38.
cle. In: 2019 IEEE international conference on environment and Zhang J, Wu WJ, Xie WB, et al. (2021) Dimensional-varying integral
electrical engineering and 2019 IEEE industrial and commercial sliding mode controller design for uncertain Takagi–Sugeno fuzzy
power systems Europe (EEEIC / I&CPS Europe), Genova, 11– systems. Information Sciences 565: 77–90.
14 June. Zhang Q, Li Y, Li C, et al. (2020) Real-time adjustment of load fre-
Sharma V and Purwar S (2014) Nonlinear controllers for a light- quency control based on controllable energy of electric vehicles.
weighted all-electric vehicle using Chebyshev neural network. Transactions of the Institute of Measurement and Control 42(1):
International Journal of Vehicular Technology 2014: 867209. 42–54.
Singh KV, Bansal HO and Singh D (2020) Feed-forward modeling and Zhang X, Wang Y, Liu G, et al. (2016) Robust regenerative charging
real-time implementation of an intelligent fuzzy logic-based energy control based on T–S fuzzy sliding-mode approach for advanced
management strategy in a series–parallel hybrid electric vehicle to electric vehicle. IEEE Transactions on Transportation Electrifica-
improve fuel economy. Electrical Engineering 102(2): 967–987. tion 2(1): 52–65.
Sun C, Gong G, Yang H, et al. (2019) Fuzzy sliding mode control for Zhu Y, Zhao H, Sun H, et al. (2020) Robust control design of electric
synchronization of multiple induction motors drive. Transactions helicopter tail reduction system: Fuzzy and optimal view. Journal
of the Institute of Measurement and Control 41(11): 3223–3234. of Vibration and Control 26(9–10): 814–829.

View publication stats