Received April 20, 2021, accepted May 8, 2021, date of publication May 13, 2021, date of current version

May 24, 2021.

Digital Object Identifier 10.1109/ACCESS.2021.3080086

Electronically Tunable ACO Based Fuzzy

FOPID Controller for Effective Speed
Control of Electric Vehicle
MARY ANN GEORGE 1 , (Graduate Student Member, IEEE),
DATTAGURU V. KAMAT 1 , (Senior Member, IEEE), AND
CIJI PEARL KURIAN 2 , (Senior Member, IEEE)
1 Department of Electronics and Communication Engineering, Manipal Institute of Technology, Manipal Academy of Higher Education (MAHE),
Manipal 576104, India
2 Department of Electrical and Electronics Engineering, Manipal Institute of Technology, Manipal Academy of Higher Education (MAHE), Manipal 576104, India

Corresponding author: Dattaguru V. Kamat ([email protected])

ABSTRACT The phenomenal growth of the Electric Vehicle (EV) technology demands efficient and
intelligent control strategies for the propulsion system. In this work, a novel fuzzy fractional order PID
(FOPID) controller using Ant Colony Optimization (ACO) algorithm has been proposed to control EV speed
effectively. The controller parameters and the fuzzy logic controller’s membership functions are tuned and
updated in real-time using the multi-objective ACO technique. The proposed controller’s speed tracking
performance is verified using the new European driving cycle (NEDC) test in the MATLAB-Simulink
platform. The proposed controller outperforms the ACO-based fuzzy integer-order PID (IOPID), FOPID, and
traditional IOPID controllers. The sensitivity analysis confirms the robustness of the proposed controller for
varying parameters of the EV model. The stabilization of EV speed in the presence of external disturbance is
also confirmed. In the proposed work, an attempt is made to analyze the system’s stability using Matignon’s
theorem, considering the linearized EV model. The proposed controller gives optimum speed tracking
performance compared to the Genetic Algorithm (GA) and the Particle Swarm Optimization (PSO) based
fuzzy FOPID controllers. Additionally, the optimized fuzzy FOPID controller is realized using a second-
generation current conveyor with extra inputs (EX-CCII) and fractional-order capacitors with electronic
tunability. The controller circuit’s performance evaluation is carried out in the Cadence Analog Design
Environment using GPDK 180 nm CMOS process.

INDEX TERMS Ant colony optimization, electric vehicle, multi-objective optimization, fuzzy FOPID,
second-generation current conveyor with extra inputs.

I. INTRODUCTION of the EV [3]. Efficient performance and desirable energy

The rise in environmental concerns and demand for fossil fuel management are the two key parameters that require intensive
resources has necessitated incorporating electric vehicle (EV) and focused investigations. The controller should provide
technology. In the recent past, EVs have gained popularity the maximum speed with low tracking error and energy
concerning their high efficiency, low maintenance cost, and consumption [4]. The EV system is highly non-linear, time-
easy operations [1], [2]. The emerging trend in EVs has dependent, and uncertain due to the varying road conditions,
led to massive pollution reduction and better sustainability motor parameters, and external disturbances. Hence, design-
in urban cities. The propulsion system has been an inte- ing a controller that eliminates the external disturbances and
gral part in deciding the overall performance of EV. The handling uncertainties with low control signal has become a
researchers at industrial and academic levels have primarily challenge [5].
focused on developing controls for the propulsion system The conventional PID controllers are generally used in
various industrial applications due to their simplicity and ease
The associate editor coordinating the review of this manuscript and of tuning [6], [7]. However, they do not perform effectively
approving it for publication was Haibin Sun . at varied operating conditions and do not assure desired

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
73392 VOLUME 9, 2021
M. A. George et al.: Electronically Tunable ACO Based Fuzzy FOPID Controller for Effective Speed Control of EV

dynamic performance [8]. The use of fuzzy logic control self-tuning feature can even enhance the controller capability
with PID controllers enhances the classical PID controller’s and system performance.
performance with self-tuning features [9]–[13]. Fuzzy con- The majority of the controllers in current industries have
trollers have been widely used in controlling EV systems. been implemented in the digital form using PLC or micro-
Khatun et al. [14] developed a fuzzy controller to control the processors. However, the digital controllers have low speed
EV antilock braking system by compensating for the non- and low memory capacity, making them unsuitable for fast
linear dynamics. A fault-tolerant fuzzy controller can raise processes such as speed control of EVs and chemical reac-
EV’s initial torque with variable characteristics of speed and tions [32]. The digital implementation also suffers from high
high efficiency [15]. power consumption related to the analog-to-digital (A/D)
The emergence of fractional calculus has led to the devel- converter.
opment of fractional order PID controller that offers two There have been several works on the analog circuit
additional degrees of freedom, the non-integer order of the realization of the FOPID controller reported in the litera-
integrator and the differentiator stages [16]–[18]. The non- ture, using analog blocks like Operational Transconductance
integer order controller provided better servo, regulatory Amplifier (OTA) [32], [33], CCII [34], Voltage Differencing
performance, and robustness compared to its integer-order Current Conveyor (VDCC) [35]. Most of these circuits suffer
counterparts. The significant benefits of fractional order con- from drawbacks, such as a high number of active/ passive
trollers are their efficacy, flexibility in system modeling, and elements [32], [33], and lack of electronic tunability [34].
design performance [19], [20].
The artificial intelligence (AI) based controllers have A. MOTIVATION AND RESEARCH GAP
gained importance due to their satisfactory performance The majority of the reported work on tuning the fuzzy
in various motor control applications, including speed logic input and output scaling factors focus on GA [23],
assessment and torque ripple minimization [21]. However, PSO [23], [24], and Cuckoo algorithm [36]. Apart from
AI-based controllers suffer from drawbacks, such as large the scaling factors, the position of the input and output
data requirements, extended learning, and training dura- membership functions plays a vital role in the fuzzy logic
tion. A fuzzy logic controller is a powerful tool that can controller [37]. Hence, it is worth noting that the tuning of
integrate human reasoning into the controller design [13]. membership function can significantly enhance the system’s
The fuzzy controllers can operate in linear and non-linear performance.
systems without considering their accurate mathematical The Ant Colony Optimization (ACO) algorithm is pre-
models [22]. The fuzzy controllers outperform other con- ferred to optimize the controller parameters and tune the
trollers in complex and non-linear systems for which good membership functions due to its numerous advantages
practical knowledge exists. The accuracy of fuzzy logic con- compared to other optimization algorithms such as GA
trollers depends on the type and number of fuzzy member- and PSO [38]. The ACO algorithm is a meta-heuristic
ship functions and fuzzy rules. At present, the optimization approach that offers high robustness, better reliability, greater
techniques explored with fuzzy logic control have gained flexibility, fast convergence, easy implementation, and fewer
massive attention in various industrial applications due to optimization parameters [38]–[41]. It is also capable of com-
their high-quality results, high efficiency, ability to adapt, bining with other algorithms. It is well suited for feature
and high accuracy. Hence, an optimal fuzzy logic controller selection and parameter tuning with better global search abil-
can be designed by utilizing optimization techniques such ity. It is suitable for dynamic applications and can quickly
as Genetic Algorithm (GA) [23], Particle Swarm Optimiza- adapt to changes.
tion (PSO) [24], Backtracking Search Algorithm (BSA) [25], The analog circuit realization of the FOPID controller
Bee Colony Optimization (BCA) and differential evolution involves the realization of fractional-order capacitors, which
(DE) [26]. are not yet available commercially. The behavior of the
Das et al. [27] presented a GA-based optimized fuzzy fractional-order capacitors can be emulated using the
FOPID controller, which could provide a better set-point RC ladder/tree structures [42] and multiple-loop-feedback
tracking with a significant compromise in rejecting the load structures [43]. Considering the RC structures’ greater energy
disturbance. Kumar et al. [28] investigated the design of a consumption and a more significant number of active element
cascade fractional-order fuzzy PI and PD controller for a count required for the multiple-loop-feedback structures, a
hybrid electric vehicle based on a multi-objective genetic better solution using a resistor less and energy-effective
algorithm. The fuzzy FOPID controllers have been widely structure to realize the fractional-order capacitors is deemed
used in various applications such as vibration isolation struc- necessary.
ture [29], pneumatic pressure system [30], pumped storage
unit regulating system [31], and Automatic Generation Con- B. CONTRIBUTION AND PAPER ORGANIZATION
trol (AGC) for electrical power systems [8], [13]. It is evident 1. This work focuses on the efficient design and circuit
from the literature that combining fuzzy logic with fractional realization of a fuzzy FOPID controller for EV speed
operators could further improve the feedback control sys- control. The optimization of the input /output scaling fac-
tem’s robustness. Additionally, introducing an adaptive or tors, antecedent part of input membership function, and

VOLUME 9, 2021 73393

 M. A. George et al.: Electronically Tunable ACO Based Fuzzy FOPID Controller for Effective Speed Control of EV

coefficients of the consequent parts of the Takagi-Sugeno

fuzzy inference system is performed using the
ACO algorithm.
2. The proposed optimization is expected to minimize the
multi-objective function to improve the time-domain
performance indices. The novel controller’s ability to
reject disturbances and provide robustness to uncertain-
ties and parameter variations has also been investigated
in this study. The controller facilitates the fastest tracking
with minimum overshoot and low values of time-domain
performance indices.
3. The stability analysis and eigenvalue analysis of the
proposed ACO-based fuzzy controller and EV model is
FIGURE 1. External forces acting on a running EV.
carried out.
4. The performance of the proposed controller is also com-
pared with GA and PSO-based fuzzy FOPID controllers. where m is the mass of the electric vehicle, g is the gravity
5. The suggested controller is realized using a single acceleration, v the driving velocity of the vehicle, µrr the
EX-CCII, which provides a simultaneous realization rolling resistance coefficient, ρ the air density, A the frontal
of the fractional-order integrator and the differentiator area of the vehicle, Cd the drag coefficient and ϕ the hill-
stages of various orders and the unity gain frequencies. climbing angle. Table 1 describes the EV parameters and
An OTA-based resistorless topology is employed to emu- specifications.
late the fractional-order capacitors used to realize the
fractional-order differentiator and the integrator stages. TABLE 1. EV parameters and specifications [45].
This study anticipates gaining some valuable and novel
insights into the effective real-time performance of the
EV propulsion system to find broad applications in the
ongoing efforts in sustainable growth.
The paper has been organized as follows: Section II describes
the mathematical model of the EV. Section III gives the
fuzzy FOPID controller structure with details of the
two-dimensional rule base and membership functions. It also
explains the formulation of a multi-objective function along
with the ACO algorithm used for optimization. The circuit
realization of the fuzzy FOPID controller using the EX-CCII
with electronic tuning features is presented in Section IV.
Section V compares the performances of the fuzzy IOPID The resultant force Ft produces a torque TL to the driving
and the fuzzy FOPID controllers for set-point tracking, distur- motor and is given by
bance rejection, and uncertainties. It also presents the results r
of circuit simulation, and the concluding remarks are outlined TL = Ft × (2)
G
in Section VI. where r is the EV tire radius and G the gearing ratio.
II. MATHEMATICAL MODEL OF ELECTRIC VEHICLE The non-linear model of the DC motor [44] is given by
The EV mainly comprises a battery unit, controller, and di 1 

electric motors connected to the vehicle through the trans- = V − Ra + Rf i − Laf iω (3a)
dt (La + Lf )
mission unit. The EV system dynamics has two parts: vehicle dω 1n o
and motor dynamics. The electric vehicle system modeling = Laf i2 − Bω − TL (3b)
dt J
involves the balancing of all the forces acting on a run-
where i is considered the armature and field current, ω the
ning vehicle. There are mainly four types of forces, namely
angular speed of the motor, La the armature inductance,
rolling friction (Frr ), aerodynamic drag force (Fad ), gravita-
Ra the armature resistance, Lf the field winding inductance,
tional force (Fg ), and force due to vehicle acceleration (Fa ),
Rf the field winding resistance, Laf the mutual inductance
as shown in Figure 1.
among the field and armature windings, B the viscous coef-
Hence, the total traction force (Ft ) acting on a vehicle is
ficient, J the moment of inertia of the motor, TL the external
given by
torque and V the input voltage.
Ft = Frr + Fad + Fg + Fa (1a) Hence, the driving velocity of the vehicle v is given by
dv r
= µrr mg + 0.5ρACd v2 + mgsinϕ + m (1b) v=ω× (4)
dt G
73394 VOLUME 9, 2021
M. A. George et al.: Electronically Tunable ACO Based Fuzzy FOPID Controller for Effective Speed Control of EV

Therefore, by combining the vehicle and the motor dynam- Fractional calculus is an essential branch of mathematics that
ics, the overall EV model is given by uses non-integer order powers of integration and differential
di 1 
operators.
= V − Ra + Rf i − Laf iω (5a) The differ-integration operator α Drt represents a fractional
dt (La + Lf )
order differentiation and integration as in (7)
dω 1 n r
= Laf i2 − Bω − (µrr mg
dt (J + m(r/G) ) 2 G
 dr 

 r r > 0 

dt
o
2 r
+0.5ρACd v + mgsinϕ (5b) α Dt 1r = 0  (7)

Rt
α (dτ )
−r
r <0

The equations (5a) and (5b) of the EV can be represented in
Simulink, as shown in Figure 2. where r ∈ R is the order of the operation and α, t the lower
and the upper limits.
Several definitions have been reported in the literature
to define the differ-integration operator, such as Reimann-
Liouville, Grunwald-Letnikov, Caputo, Cauchy integral for-
mula. The fractional-order operator sr can be approximated to
an integer order rational function using Oustaloup’s approxi-
mation method [47]. Oustaloup’s method is based on a recur-
sive distribution of poles and zeros for a frequency range
of [ωb , ωh ]. Oustaloup’s approximation for the analog filter
takes the form
N
Y s + ωk0
sr ∼
=C (8a)
s + ωk
k=−N

where r ∈ [−1, 1] ⊆ R
The expressions for zeros, poles and gain are given by
 k+N +0.5(1−r)
ωh

2N +1
ωk0 = ωb
ωb
 k+N +0.5(1+r)
ωh

2N +1
ωk = ωb (8b)
ωb
C = ωhr (8c)
FIGURE 2. Representation of an EV system in Simulink.
Here, ωb is the lower transitional frequency, and ωh is the
The non-linear model in (5) can be converted into state- higher transitional frequency. The unity gain frequency ωo is
√
space form as calculated by ωo = ωb ωh and order of the transfer function
is n = 2N +1, which can only be an odd-order approximation.
Ẋ = f (X ) + g(X )u (6) By selecting N = 2 and the frequency band as [10−3 , 103 ],
where the analog filter order turns out to be equal to 5.
    The expression of the FOPID controller is given as
x1 i
X = =
x2 ω Ki
C(s) = Kp + + Kd sµ (9)
Ra + Rf Laf sλ
 
− x1 − x1 x2

 La + Lf La + Lf 
 where Kp is the proportional gain, Ki the integral gain, Kd the
1 n r derivative gain, λ the order of the integrator stage, and µ the
 J + m(r 2 /G2 ) Laf x1 − Bx2 − G (µrr mg
2
f (X ) = 
 

 order of the differentiator stage. The time-domain expression
r2 2
  
1 of the control output of the FOPID controller is given by
+ ρACd 2 x2 + mgsinϕ)
 
2 G
" #
1 u(t) = Kp e(t) + Ki D−λ e(t) + Kd Dµ e(t) (10)
g(X ) = La +Lf , h(X ) = x2
0 where e(t) is the tracking speed error.
The structure of a fuzzy FOPID controller is shown
III. DESIGN OF A FUZZY FRACTIONAL ORDER PID in Figure 3. The error (e) and the fractional derivative of
During the last few decades, fractional calculus has error (de) are the two inputs to the fuzzy FOPID, and o is
been widely used in solving control problems [46]. the output of the fuzzy FOPID controller.

VOLUME 9, 2021 73395

 M. A. George et al.: Electronically Tunable ACO Based Fuzzy FOPID Controller for Effective Speed Control of EV

Here, a Takagi-Sugeno type fuzzy inference system (FIS)

is used. The FIS has three blocks, i.e., fuzzification,
decision-making logic with rule base, and defuzzification,
as shown in Figure 4. In the fuzzification stage, the crisp input
values are converted to a linguistic variable using a triangular
membership function, with a 50% overlap. The triangular
FIGURE 3. Structure of fuzzy FOPID controller.
membership function is described as


 0, x≤0
x − ak


, ak ≤ x ≤ bk

By using linear transformation for the control output u(t)


in Figure 3, we get f (x, a, b, c) = bckk −−x
ak (13)

 , bk ≤ x ≤ ck
 ck − bk
u(t) = Ke e(t) + Kde Dµ e(t) + D−λ e(t) Ku



(11) 
 0, ck ≤ x
By comparing (10) and (11), the gain expressions can be where ak , ck denote the feet and bk the peak of the tri-
given as angular membership function. The distribution of member-
ship functions for the input variables e and de are shown
Kp = Ke Ku (12a) in Figures 5 (a) and 5 (b). The input variables have five
Kd = Kde Ku (12b) fuzzy sets: Negative Big (NB), Negative Medium (NM),
Ki = Ku (12c) Zero (Z), Positive Medium (PM), and Positive Big (PB).
Figure 6 shows the distribution of the output membership
where Ke , Kde are the input scaling factors and Ku the output function. The fuzzy IF-THEN rule describes a condition
scaling factor. that relates the linguistic variables and fuzzy sets to the

FIGURE 4. Block diagram of the proposed optimal fuzzy FOPID controller scheme for EV speed control.

73396 VOLUME 9, 2021

M. A. George et al.: Electronically Tunable ACO Based Fuzzy FOPID Controller for Effective Speed Control of EV

TABLE 2. Rule base for fuzzy FOPID controller.

A. ANT COLONY OPTIMIZATION (ACO)

The ACO is one of the robust and adaptive algorithms used
to solve optimization problems based on the natural behavior
of the ants [49]. The optimal solution can be determined
when the ants’ colony communicates with each other using
an indirect method called the pheromone decomposition. The
shortest distance from the initial state to the destination is
found using a sequence of neighboring states. This algorithm
can find the optimal solution faster when a higher number
of pheromones are released. The pheromone matrix, which
is used to determine the optimal solution, is ψ = ψab . The
initial state of the pheromone matrix is given by
ψab = ψ0 ∀(a, b) (14)
FIGURE 5. Distribution of input membership function (a) error (e),
(b) fractional derivative of error (de). where ψ0 > 0. The probability (PYa.b ) of selecting node a at
node b is given as
[ψab (t)]α [ηab ]β
PYa.b = P α β
(15)
a,b∈T y [ψab (t)] [ηab ]

where T y defines the path executed at a given time by an

ant (Y ), α and β are the constants that determine the relative
impact of the pheromones and the heuristic factors on ants’
decision. The heuristic factor ηab is given by
1
ηab = (16)
distance between nodes a and b
FIGURE 6. Distribution of the output of Takagi-Sugeno type FIS.
The quality of pheromone 1ψab
Y at each path is defined as
 
output [48]. Table 2 describes the 25 IF-THEN rules used in L best
this work. The Takagi-Sugeno type FIS gives a crisp output, 1ψab
Y
=  LY  (17)
either a linear combination of the inputs or a constant. Hence, 0
it is considered as a weighted average defuzzification process. where L best is the best solution in the current iteration and
The proposed scheme of the fuzzy FOPID controller for the L Y is the value of the objective function determined by an
EV system is illustrated in Figure 4. The Takagi-Sugeno FIS ant (Y ).
consumes less time compared to that of a Mamdani fuzzy A phenomenon known as pheromone evaporation is
system [9]. adopted to delete the previous pheromones when a better
The input and output scaling factors (Ke , Kde , Ku ), optimal solution is reached.
adjustable parameters of input membership function (X , Y ), The expression for pheromone evaporation is given as
coefficient of the consequent part (Z ), the order of
NY
integrator (λ), and order of differentiator (µ) are varied to X
achieve an optimal solution and improve the speed tracking ψab (t) = ρψab (t − 1) + 1ψab
Y
(t) (18)
performance of electric vehicle system. Y =1

The fuzzy FOPID controller is tuned using the ACO algo- where ρ (0 < ρ ≤ 1) is the evaporation rate, and NY denotes
rithm, and its performance is compared with other optimiza- the number of ants. Figure 7 shows the pseudocode for the
tion algorithms in the MATLAB-Simulink platform. ACO algorithm.

VOLUME 9, 2021 73397

 M. A. George et al.: Electronically Tunable ACO Based Fuzzy FOPID Controller for Effective Speed Control of EV

IV. REALIZATION OF FOPID CIRCUIT USING EX-CCII

The optimum fractional PID controller can be real-
ized using the extra-X second-generation current conveyor
(EX-CCII) [34]. The main advantage of this structure is that
a single active element is used to realize the controller, and
the fractional-order differentiator and the integrator stages of
any order can be implemented using the structure, as shown
in Figure 8.

FIGURE 7. Pseudocode for ACO algorithm.

B. FORMULATION OF OBJECTIVE FUNCTION FOR TUNING

FIGURE 8. Realization of the FOPID controller using EX-CCII [34].
A controller can be optimal when its control parameters are
adjusted such that the cost function is minimized. In multi-
objective optimization, the cost function is a weighted sum The terminal properties of EX-CCII are given by
of two or more objective functions. During optimization, VX 1 = VX 2 = VX 3 = VY
it is crucial to minimize both the error index and the control
signal. This optimization type can reduce the control sig- iZ 1 = iX 1 , iZ 2 = iX 2 , iZ 3 = iX 3
nal’s value, preventing the actuator’s integral wind-up and RY → ∞ (20)
saturation. In this study, five performance indices have been
where VXk (k = 1, 2, 3) are the voltages at input terminals
considered as follows:
Z ∞ Xk, VY is the voltage at terminal Y , iZk (k = 1, 2, 3) are the
J1 = ITSE + ISCO = te2 (t)dt + ISCO (19a) currents at terminal Zk and iXk (k = 1, 2, 3) are the currents
Z0 ∞ at terminals Xk.
The FOPID controller expression derived from applying
J2 = ITAE + ISCO = t |e(t)| dt + ISCO (19b)
0 terminal properties of EX-CCII in Figure 8 is given as
Z ∞
R2 1
J3 = IAE + ISCO = |e(t)| dt + ISCO (19c) C(s) = + + Rµ Cµ sµ (21)
0
Z ∞ R1 Rλ Cλ sλ
J4 = ISE + ISCO = e2 (t)dt + ISCO (19d) Here, Cλ and Cµ are the pseudo-capacitance with units
0
Farad/sec1−λ and Farad/sec1−µ .
J5 = ITSE + ITAE + IAE + ISE + ISCO (19e)
Z ∞ By comparing (21) and (9), we get
ISCO = u2 (t)dt (19f)
R2 1
0 Kp = , Ki = , Kd = Rµ Cµ (22)
where e(t) is the error signal, u(t) is the control signal, ITSE R1 Rλ Cλ
is the integral time square error, ITAE is the integral time The fractional-order capacitors are approximated using the
absolute error, IAE is the integral absolute error, ISE is the modified Oustaloup’s approximation and realized using the
integral square error, and ISCO is the integral of the RC Valsa network, as shown in Figure 9.
squared control signal. Each of these performance indices The details of the multi-functional EX-CCII analog block
has certain advantages in the control system design [50]. and the three-input summation stage have been described
These performance indices are considered as the objec- in [34]. Figure 10 illustrates the CMOS realization of the
tive function for tuning, ensuring stability and better three-input EX-CCII and the three-input summation stage.
speed tracking performance when there is sudden load The EX-CCII circuit provides an accurate voltage conveying
disturbance, parameter variation, and reference speed from terminal Y to terminals X1, X2, and X3. The currents
variation. from terminals X1, X2, and X3 are copied to terminals
The ACO algorithm minimizes the objective function Ji Z1, Z2, and Z3, respectively. The minimum supply voltage
(i = 1, 2, 3, 4, 5) to produce the optimally tuned required is VTHn +2VDS,sat.
input and output scaling factors, integral-differential orders, As the EX-CCII analog blocks are not available com-
and adjustable parameters of membership functions of the mercially, the FOPID controller circuit can be realized
fuzzy FOPID controller with a low control signal and using CCII/ CFOA integrated circuit (IC) AD844, as shown
error-index. in Figure 11.

73398 VOLUME 9, 2021

M. A. George et al.: Electronically Tunable ACO Based Fuzzy FOPID Controller for Effective Speed Control of EV

FIGURE 9. Valsa RC network.

FIGURE 11. FOPID controller circuit using IC AD844 (current feedback

operational amplifier).

FIGURE 10. CMOS circuit of (a) three input EX-CCII, (b) three input
summation stage [34].

The output expression for the summation stage, shown

in Figure 10 (b), is given as
Vout = gm R(Vin1 + Vin2 + Vin3 ) (23)
Here, gm is the transconductance of the transistors Mn1a-
Mn6a and the resistance R = 1/gm .
The electronic tunability of the EX-CCII based FOPID
controller circuit in Figure 8 can be achieved by replac-
ing all the passive grounded and floating resistors using
operational transconductance amplifier (OTA) simulated
resistors [51], [52], as shown in Figure 12 (a) and (b), respec-
tively. This circuit offers benefits such as electronic tunability,
wide bandwidth, simple design, and a wide range of resis-
tance between 50 M and 1 k. FIGURE 12. CMOS circuit of an electronically tunable resistor using OTA
(a) grounded type, (b) floating type.
Assuming matched transistors, in Figure 12, the expression
for current I1 is given by
where V1 , V2 are the input voltages of the transconductance
I1 = −I 2 = Gm (V1 − V2 ) (24) amplifier and Gm is the transconductance.

VOLUME 9, 2021 73399

 M. A. George et al.: Electronically Tunable ACO Based Fuzzy FOPID Controller for Effective Speed Control of EV

FIGURE 13. Output surface of fuzzy FOPID and IOPID controllers after ACO using J1 , J2 , J3 , J4 , and J5 objective functions.

The resistance R and R12 can be found as TABLE 3. Parameters for ACO.

V1 1
R= =√ (25a)
I1 βIBias
V1 − V2 V2 − V1 1
R12 = = =√ (25b)
I1 I2 βIBias
where IBias is the input biasing current β is the transconduc-
tance parameter of the MOS differential pair.

V. RESULTS AND DISCUSSION

The optimal fuzzy FOPID controller for EV speed control
shown in Fig. 4 is simulated using MATLAB-Simulink and
Cadence Virtuoso Analog Design Environment. This section
demonstrates the superiority of the ACO-based fuzzy FOPID
controller over fuzzy IOPID controller, FOPID, and con- the optimization. The blue and orange colour plots represent
ventional IOPID controller through simulation of the EV the output surface plot of fuzzy IOPID and fuzzy FOPID,
system’s speed tracking performance. respectively.
The parameters selected for ACO are given in Table 3. Four operating scenarios are considered to validate the
The ACO minimizes the objective function (19a)-(19e) to effectiveness and robustness of the proposed controllers,
determine the fuzzy controllers’ optimal control parameters. namely set-point tracking, disturbance rejection, noise sup-
Table 4 shows the fuzzy FOPID and fuzzy IOPID controller pression, and sensitivity analysis. This section also presents
parameters obtained after optimization considering various the stability analysis and eigenvalue analysis of the EV sys-
objective functions. Here, X, Y are the antecedent values, and tem. The performance comparison of the proposed scheme
Z the value of the consequent. with other existing controllers is also described in this section.
Figure 13 illustrates the non-linear surface plot of the fuzzy a. Set-Point Tracking: The New European Driving
IOPID and fuzzy FOPID obtained after ACO by minimizing Cycle (NEDC) test is performed to validate the fuzzy FOPID
the J1 , J2 , J3 , J4 , J5 objective functions. It shows the input controller’s performance. The NEDC has been commonly
and output relationship of the fuzzy logic controller. Here, used to test the light-weighted EVs in Europe and India [53].
the three axes are the error (e), the fractional derivative of The maximum speed of the NEDC cycle is 120 km/h,
error (de), and the output (o) of the Takagi-Sugeno FIS. as shown in Figure 14. The proposed fuzzy FOPID controller
It illustrates that the distribution of e, de, and the coef- and fuzzy IOPID controller’s performance to track the NEDC
ficient of the consequent part of the output varies during test is compared and illustrated in Figure 15 (a). It also shows

73400 VOLUME 9, 2021

M. A. George et al.: Electronically Tunable ACO Based Fuzzy FOPID Controller for Effective Speed Control of EV

TABLE 4. Optimal set of tuning parameters for fuzzy FOPID and fuzzy IOPID controllers with ACO.

TABLE 5. Performance parameters for fuzzy FOPID and IOPID controllers.

the speed tracking performance of an ACO-based IOPID and controllers have superior performance compared to J1 , J2 , J3,
FOPID controller. The proposed controllers’ effectiveness and J4 optimized controllers.
is demonstrated by plotting the error signals and controller The multi-objective optimization can result in solutions
effort for each controller, as shown in Figure 15 (b) and 15 (c), called the Pareto optimal solutions or non-dominant solu-
respectively. As it can be inferred, while the IOPID controller tions. Figure 16 shows the distribution of the non-dominant
produces the highest control effort and error signal, fuzzy solutions in the 4-dimensional Pareto optimal front (ITAE,
FOPID generates the lowest control effort and error signal IAE, ITSE, ISE) using multi-objective ACO. Here, J5 multi-
making its performance superior to others. objective function is chosen, and the resulting convergence
Table 5 summarizes the performance parameters of fuzzy graph of multi-objective ACO for 100 generations is illus-
FOPID and fuzzy IOPID controllers for various objective trated in Figure 17.
functions. The time-domain specifications such as settling b. Disturbance Rejection: The robustness and the effec-
time, rise time, percentage overshoot, steady-state error, and tiveness of the fuzzy FOPID controller are verified in actual
the performance indices such as ITSE, ITAE, IAE, ISE, and working conditions by introducing disturbances. An efficient
J5 are compared for both controllers. Critical examination and robust controller must reject the disturbance such that the
reveals that the fuzzy FOPID controllers’ performance is deviation from the desired response is minimum. The speed
far better than the fuzzy IOPID controllers with high accu- tracking performance of the suggested controllers under the
racy, less settling time, percentage overshoot, steady-state influence of disturbance is shown in Figure 18. The results
error, and error indices. It also shows that the J5 optimized show that the ACO-based IOPID and the FOPID cannot

VOLUME 9, 2021 73401

 M. A. George et al.: Electronically Tunable ACO Based Fuzzy FOPID Controller for Effective Speed Control of EV

FIGURE 14. New European drive cycle.

accurately track the NEDC cycle than the fuzzy-based TABLE 6. Uncertain parameters Of EV system.
controllers. The fuzzy-based controllers can return to the
set-point value quickly after the appearance of external dis-
turbance. Also, such a system requires less recovery time
compared to others.
c. Noise suppression: The EV system’s robustness in the
presence of measurement noise is tested by introducing a
random signal of amplitude -0.04 to +0.04 and sampling
time 0.01 seconds. Figure 19 demonstrates the effects of
adding the noise input to the system. The fuzzy FOPID gives
relatively minor fluctuation than fuzzy IOPID, FOPID, and
IOPID controllers, showing a superior and robust control
performance in noise suppression.
d. Sensitivity Analysis and Robustness: The controllers’ where G(s) is the plant transfer function, C(s) indicates the
robustness is demonstrated by introducing uncertainties and controller transfer function, and L(s) = G(s)C(s) represents
varying EV system parameters. Here, the uncertain parame- the loop transfer function.
ters of the EV like mass (m), drag coefficient (Cd ), rolling The sensitivity function shows the system’s ability
resistance coefficient (µrr ) and EV tire radius (r) are var- to suppress load disturbances and attain good set-point
ied, and the percentage of variation in these parameters is tracking. The complementary sensitivity function specifies
shown in Table 6. Figure 20 shows the robustness of the the robustness against the measurement noise [54]. The
suggested controllers against the variations in system param- frequency-domain plots of sensitivity function, complemen-
eters, i.e., change in m by +30%, µrr by +30%, Cd by tary sensitivity function, disturbance sensitivity, and control
−20% and r by +25%. It is observed that, compared to other sensitivity are shown in Figure 21. For satisfactory system
controllers, the fuzzy FOPID controller takes the minimum performance, the sensitivity function must have a small value
time to complete the full power acceleration and stabilize in at lower frequencies, and the complementary sensitivity func-
the presence of the uncertainties. tion must have a small value at higher frequencies. The plots
The critical frequency domain specifications are [54]: show that the fuzzy FOPID controller provides a better load
Sensitivity function disturbance rejection and a better high-frequency measure-
ment noise rejection than other controllers. It is also observed
1 that the sensitivity peak under fuzzy FOPID controller is min-
S(s) = (26a)
1 + L(s) imum, while the conventional IOPID and FOPID controllers
Complementary Sensitivity function have higher sensitivity peaks.
L(s) e. Matignon’s Theorem and Stability Analysis:
T (s) = (26b) Theorem: The fractional-order transfer function G(s) =
1 + L(s) 
N (s) D(s) is stable in s-plane if and only if the following
Disturbance Sensitivity condition is satisfied [55]:
G(s)
Sd (s) = (26c)
1 + L(s) π
|arg(wi )| > q , ∀wi ∈ C, (27)
Control Sensitivity 2
C(s)
Su (s) = (26d)
1 + L(s) the ith root of D(w) = 0, where w = sq , (0 < q < 2).
73402 VOLUME 9, 2021
M. A. George et al.: Electronically Tunable ACO Based Fuzzy FOPID Controller for Effective Speed Control of EV

FIGURE 15. Performance of fuzzy FOPID, fuzzy IOPID, FOPID, and IOPID (a) to track NEDC speed test, (b) error signal,
(c) controller effort.

VOLUME 9, 2021 73403

 M. A. George et al.: Electronically Tunable ACO Based Fuzzy FOPID Controller for Effective Speed Control of EV

Here, one set of the ACO based fuzzy controller parameters

is considered, i.e., Ku = 23.15, Ke = 1.69, Kce = 13.78,
λ = 0.514 and µ = 0.902.
Hence, the expression of the FOPID controller takes the
form
23.15
C(s) = 39.12 + 0.514 + 319s0.902 (29)
s
Hence, the characteristic equation of the system is given as
1 + G (s) C (s) = 0
s5.514 + 0.2985s4.514 + 4.1151s4.416 + 0.61855s3.514
+ 1.8961s3.416 + 0.29863s3 + 0.24785s2.514
+ 0.12868s2.416 + 0.1376s2 + 0.017162s1.514
+ 0.0058568s1.416 + 0.0093387s + 0.00076465s0.514
+ 0.00042503 = 0 (30)
This can be rewritten as
551.4 451.4 441.6
D(s) = s 100 + 0.2985s 100 + 4.1151s 100
351.4 341.6 300
+ 0.61855s 100 + 1.8961s 100 + 0.29863s 100
251.4 241.6 200
+0.24785s 100 + 0.12868s 100 + 0.1376s 100
151.4 141.6 100
+0.017162s 100 +0.0058568s 100 +0.0093387s 100
51.4
+ 0.00076465s 100 + 0.00042503 = 0 (31)
FIGURE 16. (a) 4D Pareto front using multi-objective ACO, (b) Zoomed
plot.
The following transformation is used to map from s-plane
to w-plane.
1
w = sm , m = 100 (32)
Therefore,
D(w) = s551.4 + 0.2985s451.4 + 4.1151s441.6
+ 0.61855s351.4 + 1.8961s341.6 + 0.29863s300
+0.24785s251.4 + 0.12868s241.6 + 0.1376s200
+ 0.017162s151.4+0.0058568s141.6+0.0093387s100
+ 0.00076465s51.4 + 0.00042503 = 0 (33)
The stability conditions for the fractional-order system are
given as
• The system is stable if
π π
< |arg(w)| < (34a)
2m m
FIGURE 17. Convergence graph of ACO. • The system is oscillatory if
π
|arg(w)| = (34b)
2m
The linearized model of the EV system, which is obtained If not, the system is unstable.
using system identification, is given by The pole-zero plot is obtained by solving (33) using the
fractional-order modeling and control (FOMCON) toolbox,
as shown in Figure 22. It shows that the system is stable
0.01292s3 + 0.005944s2
 
for q = 1/m = 0.01, and all the poles of s0.01 polynomial
+0.0004034s + 1.836e − 05
G (s) =  are placed in the stable area (outside the red shaded region),
s5 + 0.2985s4 + 0.1139s3

satisfying Matignon’s stability theorem [56]. The region of
+0.01532s2 + 0.001381s + 4.641e − 05 stability depends on the order q. Since, q = 0.01, the angle is
(28) around 0.9◦ .

73404 VOLUME 9, 2021

M. A. George et al.: Electronically Tunable ACO Based Fuzzy FOPID Controller for Effective Speed Control of EV

FIGURE 18. Performance of fuzzy FOPID, fuzzy IOPID, FOPID, and IOPID to track NEDC speed test under
the influence of disturbance.

FIGURE 19. Performance of fuzzy FOPID, fuzzy IOPID, FOPID, and IOPID to track NEDC speed test in the presence of
measurement noise.

TABLE 7. Comparison of performance of ACO, PSO, and GA based fuzzy FOPID controller for EV speed control.

Similarly, during ACO, each combination of controller stability theorem. Hence, all the controller parameter values
parameters is subjected to stability check using the Matignon that cause instability of the closed-loop system are rejected.

VOLUME 9, 2021 73405

 M. A. George et al.: Electronically Tunable ACO Based Fuzzy FOPID Controller for Effective Speed Control of EV

FIGURE 20. Robustness of fuzzy FOPID, fuzzy IOPID, FOPID, and IOPID against the parameter
variations of EV, i.e., change in mass by +30%, rolling resistance coefficient by +30%, drag
coefficient by −20%, and EV tire radius by +25%.

FIGURE 21. Frequency domain plots of (a) Sensitivity function, (b) Complementary sensitivity function, (c) Disturbance sensitivity, and (d) Control
sensitivity using fuzzy FOPID, fuzzy IOPID, FOPID, and IOPID controllers.

f. Eigenvalue Analysis: The eigenvalues of the Theorem: If all the eigenvalues of Ac satisfy the condition
compensated system can be determined using the character- qπ
istic equation given by |arg(λ(Ac ))| > (36)
2
|λI − Ac | = 0 (35) then the zero solution of the system is asymptoti-
cally stable. The proof of this theorem is detailed
where Ac is the system matrix of the linearized system with in [56].
the selected controller, λ is the eigenvalues, and I is the There are 551 roots, and all roots of the characteristic
identity matrix. equation satisfy the (36) and lie within the stable region,

73406 VOLUME 9, 2021

M. A. George et al.: Electronically Tunable ACO Based Fuzzy FOPID Controller for Effective Speed Control of EV

TABLE 8. Comparison of performance of ACO based fuzzy controller with other existing controllers.

TABLE 10. Values of resistors, DC bias currents, and capacitors of Valsa

RC networks used to realize fractional-order capacitors.

FIGURE 22. Stability plot for the closed-loop EV system.

TABLE 9. Design details for three input EX-CCII and three input
summation circuits.
iteration = 100, population size = 100, acceleration factors
c1 = c2 = 2 and inertia weights wmax = 0.9 and
wmin = 0.4. Similarly, the parameters of the GA optimization
are also selected. Here, maximum generation is taken as
100, population size = 100, crossover fraction = 0.8 and
mutation fraction = 0.2. In all the cases, J5 is considered
as the objective function to be minimized. Here, the lower
and upper bounds of the controller parameters and adjustable
membership parameters are taken from Table 3. Table 7
gives the EV time-domain performance and the performance
indices using the above-considered controllers. It is evident
from the results that the ACO-based fuzzy FOPID controller
is better than the other controllers. Also, the PSO-based
fuzzy FOPID controller gives better performance than the
GA-based controller.
Three standard error measurement criteria that can evaluate
the efficiency of the proposed controller with other existing
controllers are the sum of squared errors (SSE), mean abso-
as shown in Figure 22. This condition assures that the system lute error (MAE), and mean square error (MSE). Table 8
is bounded input bounded output (BIBO) stable and asymp- shows the performance comparison of the proposed controller
totically stable. to multi-objective PI [30], multi-objective fuzzy PI [30], and
g. Comparison of ACO Based Fuzzy FOPID Controller model predictive control (MPC) [4] controllers. It is observed
With Other Optimization Algorithms and Existing that the proposed controller yields an optimal performance
Controllers: as the values of their error measurement criteria are close to
The ACO-based fuzzy FOPID controller’s speed tracking zero.
performance is compared with the GA-based fuzzy FOPID Despite the various merits of the schemes discussed, it has
controller and the PSO-based fuzzy FOPID controller. a couple of limitations: (i) Framing the fuzzy rule base
The parameters considered for the PSO are the maximum for the fuzzy logic controller to track the new European

VOLUME 9, 2021 73407

 M. A. George et al.: Electronically Tunable ACO Based Fuzzy FOPID Controller for Effective Speed Control of EV

FIGURE 24. Frequency response of Valsa RC networks using passive

FIGURE 23. Impedance frequency response of Valsa RC networks resistors and electronically tunable resistors used to realize the
approximating the fractional-order capacitors: Cλ = 10 µ/sec0.486 fractional-order capacitors: Cλ = 10 µ/sec0.486 (a) magnitude, (b) phase
(a) magnitude (b) phase and Cµ = 10 µ/sec0.0908 (c) magnitude, and Cµ = 10 µ/sec0.0908 (c) magnitude, (d) phase.
(d) phase.

drive cycle (NEDC) test is time-consuming as it requires A. CIRCUIT REALIZATION OF FOPID CONTROLLER
expertise and experience. (ii) More number of parameters The EX-CCII based FOPID controller circuit in Figure 8
(eight parameters) are used in optimization. is simulated in the Cadence analog design environment using

73408 VOLUME 9, 2021

M. A. George et al.: Electronically Tunable ACO Based Fuzzy FOPID Controller for Effective Speed Control of EV

TABLE 11. Comparison of the proposed controller scheme with other existing solutions.

The fractional-order capacitors used in the fractional-order

differentiator and the integrator stages are approximated
using the 5th order modified Oustaloup method and realized
using the Valsa RC networks, given in Figure 9, to cover the
frequency range [0.1 Hz, 1000 Hz] with the phase accuracy
of 1◦ . The behaviour of the Valsa RC network used to imple-
ment the constant phase element is verified by plotting the
impedance frequency response, along with the ideal response,
as shown in Figure 23. The resistors in a Valsa RC network
are realized using the CMOS circuits of electronically tunable
OTA simulated resistors in Figure 12, and their values are
tuned by adjusting the bias current.
Here, the R = 1/gm , where gm is the transconductance
of the differential MOS pair in Figure 12, and the values
of DC bias currents are calculated using (25a)-(25b). Also,
the aspect ratios of the MOS transistors Mp1b-Mp4b are
set as 25 µm/2 µm and Mn1b-Mn4b as 10 µm/ 2µm.
Table 10 summarizes the value of resistors, dc bias currents,
and capacitors used in the Valsa RC network.
Figure 24 presents the magnitude and phase responses of
the Valsa RC networks using the passive resistors and the
electronically tunable resistors. It shows that the magnitude
response error is negligible, and the error in the phase plot is
FIGURE 25. Frequency response of the designed controller (a) magnitude, about 10%. These errors are caused due to the OTA’s imper-
(b) phase. fections. The gain and phase responses of EX-CCII based
FOPID controller circuits are depicted in Figure 25, which
a 180 nm GPDK CMOS process. Table 9 shows the design confirm the controller’s accurate operation. Any set of the
details and aspect ratios (W /L) for MOS transistors used controller parameters of the fuzzy FOPID controller can be
in Figure 10. realized using the circuit, shown in Figure 8, by electronically
In the CMOS circuit of three input EX-CCII, all transistors tuning the OTA simulated resistors.
operate in the saturation region, and DC bias current I0 is dis- Table 11 shows the comparison of the critical features of
tributed using the NMOS and the PMOS current mirrors with the proposed FOPID circuit to other previously reported solu-
the aspect ratios 5 µm/1µm and 25 µm/5µm, respectively. tions based on various active elements. Based on Table 11,
Similarly, the DC bias current IB in the summation stage is it is evident that in the proposed scheme, there is a signifi-
distributed using the NMOS and PMOS current mirrors with cant reduction in active element count and passive resistors.
aspect ratios 2.5 µm/1 µm and 24 µm/10 µm, respectively. OTA-based simulated resistors replace the passive resistors in
The fuzzy FOPID controller parameters in (29) are used the controller circuit and offer electronic tunability.
to evaluate the performance of the FOPID controller circuit
in Figure 8. Using Eq. (22) and (29), the circuit param- VI. CONCLUSION
eters are calculated as R1 = 1 k, R2 = 39.12 k, This study presents a novel approach in designing and devel-
Rλ = 4.319 k, Cλ = 10 µ/sec0.486 , Rµ = 31.19 M and oping a multi-objective fuzzy FOPID for speed control of EV.
Cµ = 10 µ/sec0.0908 . The EV can be controlled in real-time by adjusting the control

VOLUME 9, 2021 73409

 M. A. George et al.: Electronically Tunable ACO Based Fuzzy FOPID Controller for Effective Speed Control of EV

parameters and the membership functions via ACO when experimental setup to control EVs’ speed. The proposed
the system encounters disturbance, parameter uncertainties, fuzzy FOPID controller is well suited for cruise control appli-
and varying road conditions. The fuzzy fractional-order con- cations in EV and can also be used in EV battery recharging
trollers have become industrial control standards due to their or discharging applications under constant DC voltage. As a
improved robustness against plant parameter variations and future scope, the fuzzy-based controller can be extended to
system perturbation, and better disturbance rejection control. an Adaptive Neuro-Fuzzy Inference System (ANFIS), which
The controller’s significant advantage is its ability to reduce combines the advantages of fuzzy inference systems and
control effort, reducing the energy wasted in various indus- neural networks. It provides better learning and adaptation
trial control applications. capability without requiring expert knowledge.
The proposed controller can be effectively employed for
EV speed tracking. The effectiveness and the robustness of REFERENCES
the proposed novel controller have been comprehensively [1] M. Veysi, J. Aghaei, M. Shasadeghi, R. Razzaghi, B. Bahrani, and
illustrated by subjecting it to disturbance and uncertainties. D. J. Ryan, ‘‘Energy-efficient speed control of electric vehicles: Linear
The significant outcomes of this investigation are summa- matrix inequality approach,’’ IEEE Trans. Veh. Technol., vol. 69, no. 10,
pp. 10469–10483, Oct. 2020, doi: 10.1109/tvt.2020.3008500.
rized as: [2] M. S. Kumar and S. T. Revankar, ‘‘Development scheme and key tech-
1. The performance of ACO-based fuzzy FOPID was com- nology of an electric vehicle: An overview,’’ Renew. Sustain. Energy Rev.,
pared with the fuzzy IOPID, FOPID, and classical vol. 70, pp. 1266–1285, Apr. 2017, doi: 10.1016/j.rser.2016.12.027.
IOPID, and it was observed that the proposed controller [3] W.-J. Lee, G. Strbac, Z. Hu, Z. Ding, P. Sarikprueck, F. Teng, and
G. Kariniotakis, ‘‘Special issue on advanced approaches and appli-
gave the fastest tracking response with a settling time cations for electric vehicle charging demand management,’’ IEEE
of 0.75 sec and a rise time of 0.081 sec. The controller Trans. Ind. Appl., vol. 56, no. 5, pp. 5682–5683, Sep. 2020, doi:
exhibited a small overshoot of 0.5% and a steady-state 10.1109/tia.2020.3003567.
[4] M. H. Khooban, N. Vafamand, and T. Niknam, ‘‘T–S fuzzy model predic-
error of 0.0001. Furthermore, the proposed controller tive speed control of electrical vehicles,’’ ISA Trans., vol. 64, pp. 231–240,
gives a remarkable reduction in error indices, such as Sep. 2016, doi: 10.1016/j.isatra.2016.04.019.
IAE, ISE, ITAE, and ITSE, by 87%, 93%, 78%, and 98%, [5] M. H. Khooban, T. Niknam, F. Blaabjerg, and M. Dehghani, ‘‘Free chat-
tering hybrid sliding mode control for a class of non-linear systems:
respectively, when compared with other controllers. Electric vehicles as a case study,’’ IET Sci., Meas. Technol., vol. 10, no. 7,
2. The simulation results also revealed that the proposed pp. 776–785, Oct. 2016, doi: 10.1049/iet-smt.2016.0091.
controller could excellently handle parameter variation, [6] K. J. Åström and T. Hägglund, ‘‘The future of PID control,’’ Control Eng.
Pract., vol. 9, no. 11, pp. 1163–1175, Apr. 2001, doi: 10.1016/S0967-
uncertainties, disturbance, and noise compared to the 0661(01)00062-4.
other controllers. The proposed controller’s robustness [7] K. Heong Ang, G. Chong, and Y. Li, ‘‘PID control system analysis,
was tested under the following EV parameter variations design, and technology,’’ IEEE Trans. Control Syst. Technol., vol. 13, no. 4,
from its nominal value, i.e., change in mass +30%, pp. 559–576, Jul. 2005, doi: 10.1109/TCST.2005.847331.
[8] Y. Arya, ‘‘Impact of ultra-capacitor on automatic generation control of
change in rolling resistance +30%, change in drag coef- electric energy systems using an optimal FFOID controller,’’ Int. J. Energy
ficient -20%, and change in EV tire radius +25%. Res., pp. 8765–8778, Aug. 2019, doi: 10.1002/er.4767.
3. The stability of the system is also investigated using [9] K. Premkumar and B. V. Manikandan, ‘‘Bat algorithm optimized fuzzy
PD based speed controller for brushless direct current motor,’’ Eng.
Matignon’s stability theorem and eigenvalue analysis. Sci. Technol., Int. J., vol. 19, no. 2, pp. 818–840, Jun. 2016, doi:
4. The ACO-based fuzzy FOPID controller’s speed track- 10.1016/j.jestch.2015.11.004.
ing performance was evaluated and compared with the [10] M. Rabah, A. Rohan, Y.-J. Han, and S.-H. Kim, ‘‘Design of fuzzy-PID con-
troller for quadcopter trajectory-tracking,’’ Int. J. FUZZY Log. Intell. Syst.,
PSO and the GA optimized fuzzy FOPID controllers.
vol. 18, no. 3, pp. 204–213, Sep. 2018, doi: 10.5391/IJFIS.2018.18.3.204.
It was found that the ACO-based controller gave a faster [11] B. E. Demir, R. Bayir, and F. Duran, ‘‘Real-time trajectory tracking of
convergence and low values of performance indices, an unmanned aerial vehicle using a self-tuning fuzzy proportional integral
i.e., ITSE = 0.006, ITAE = 5.129, IAE = 0.192, ISE = derivative controller,’’ Int. J. Micro Air Vehicles, vol. 8, no. 4, pp. 252–268,
Dec. 2016, doi: 10.1177/1756829316675882.
0.03, and the sum of indices was 5.36. [12] Y. Tao, J. Zheng, Y. Lin, T. Wang, H. Xiong, G. He, and D. Xu, ‘‘Fuzzy
5. The proposed controller was realized using a sin- PID control method of deburring industrial robots,’’ J. Intell. Fuzzy Syst.,
gle EX-CCII block that offered design flexibility and vol. 29, no. 6, pp. 2447–2455, Oct. 2015, doi: 10.3233/IFS-151945.
[13] Y. Arya, ‘‘AGC performance enrichment of multi-source hydrother-
electronic tunability. It also allowed the simultaneous mal gas power systems using new optimized FOFPID controller and
realization of the fractional-order integrator and differ- redox flow batteries,’’ Energy, vol. 127, pp. 704–715, May 2017, doi:
entiator stages of different orders. This circuit can be 10.1016/j.energy.2017.03.129.
used to realize any combination of the fuzzy FOPID [14] P. Khatun, C. M. Bingham, N. Schofield, and P. H. Mellor, ‘‘Application
of fuzzy control algorithms for electric vehicle antilock braking/traction
controller parameters by adjusting the bias currents in control systems,’’ IEEE Trans. Veh. Technol., vol. 52, no. 5, pp. 1356–1364,
OTA-based resistors. The proposed circuit uses a min- Sep. 2003, doi: 10.1109/TVT.2003.815922.
imum number of passive elements that make it energy [15] S. Poiurani, K. U. Kumar, and S. Renganarayanan, ‘‘Intelligent con-
troller design for electric vehicle,’’ in Proc. 57th IEEE Semiannual Veh.
effective. The controller can be implemented using the Technol. Conf. (VTC-Spring), vol. 4, Apr. 2003, pp. 2447–2450, doi:
integrated circuits of analog blocks, and its performance 10.1109/vetecs.2003.1208830.
can be verified in real-time. [16] C. A. Monje, B. M. Vinagre, V. Feliu, and Y. Chen, ‘‘Tuning and
auto-tuning of fractional order controllers for industry applications,’’
This investigation expects to give valuable insights for future Control Eng. Pract., vol. 16, no. 7, pp. 798–812, Jul. 2008, doi:
simulation studies that can be validated using the real-time 10.1016/j.conengprac.2007.08.006.

73410 VOLUME 9, 2021

M. A. George et al.: Electronically Tunable ACO Based Fuzzy FOPID Controller for Effective Speed Control of EV

[17] F. Padula and A. Visioli, ‘‘Tuning rules for optimal PID and fractional- [35] O. Domansky, R. Sotner, L. Langhammer, J. Jerabek, C. Psychalinos, and
order PID controllers,’’ J. Process Control, vol. 21, no. 1, pp. 69–81, G. Tsirimokou, ‘‘Practical design of RC approximants of constant phase
Jan. 2011, doi: 10.1016/j.jprocont.2010.10.006. elements and their implementation in fractional-order PID regulators using
[18] A. Tepljakov, E. A. Gonzalez, E. Petlenkov, J. Belikov, C. A. Monje, CMOS voltage differencing current conveyors,’’ Circuits, Syst., Signal
and I. Petráš, ‘‘Incorporation of fractional-order dynamics into an exist- Process., vol. 38, no. 4, pp. 1520–1546, Apr. 2019, doi: 10.1007/s00034-
ing PI/PID DC motor control loop,’’ ISA Trans., vol. 60, pp. 262–273, 018-0944-z.
Jan. 2016, doi: 10.1016/j.isatra.2015.11.012. [36] R. Sharma, K. P. S. Rana, and V. Kumar, ‘‘Performance analysis of
[19] D. Guha, P. K. Roy, S. Banerjee, S. Padmanaban, F. Blaabjerg, and fractional order fuzzy PID controllers applied to a robotic manipula-
D. Chittathuru, ‘‘Small-signal stability analysis of hybrid power system tor,’’ Expert Syst. Appl., vol. 41, no. 9, pp. 4274–4289, Jul. 2014, doi:
with quasi-oppositional sine cosine algorithm optimized fractional order 10.1016/j.eswa.2013.12.030.
PID controller,’’ IEEE Access, vol. 8, pp. 155971–155986, 2020, doi: [37] C.-S. Shieh, ‘‘Fuzzy PWM based on genetic algorithm for battery
10.1109/ACCESS.2020.3018620. charging,’’ Appl. Soft Comput., vol. 21, pp. 607–616, Aug. 2014, doi:
[20] H. K, Abdulkhader, J. Jacob, and A. T. Mathew, ‘‘Robust type-2 10.1016/j.asoc.2014.04.009.
fuzzy fractional order PID controller for dynamic stability enhance- [38] A. Mughees and S. A. Mohsin, ‘‘Design and control of magnetic levitation
ment of power system having RES based microgrid penetration,’’ Int. system by optimizing fractional order PID controller using ant colony
J. Electr. Power Energy Syst., vol. 110, pp. 357–371, Sep. 2019, doi: optimization algorithm,’’ IEEE Access, vol. 8, pp. 116704–116723, 2020,
10.1016/j.ijepes.2019.03.027. doi: 10.1109/ACCESS.2020.3004025.
[21] M. A. Hannan, J. A. Ali, M. S. H. Lipu, A. Mohamed, P. J. Ker, [39] M. Birattari, P. Pellegrini, and M. Dorigo, ‘‘On the invariance of ant colony
T. M. I. Mahlia, M. Mansor, A. Hussain, K. M. Muttaqi, and Z. Y. Dong, optimization,’’ IEEE Trans. Evol. Comput., vol. 11, no. 6, pp. 732–742,
‘‘Role of optimization algorithms based fuzzy controller in achieving Dec. 2007, doi: 10.1109/TEVC.2007.892762.
induction motor performance enhancement,’’ Nature Commun., vol. 11, [40] M. Wang, T. Ma, G. Li, X. Zhai, and S. Qiao, ‘‘Ant colony optimization
no. 1, pp. 1–11, Jul. 2020, doi: 10.1038/s41467-020-17623-5. with an improved pheromone model for solving MTSP with capacity and
[22] S. Çeven, A. Albayrak, and R. Bayır, ‘‘Real-time range estimation in time window constraint,’’ IEEE Access, vol. 8, pp. 106872–106879, 2020,
electric vehicles using fuzzy logic classifier,’’ Comput. Electr. Eng., vol. 83, doi: 10.1109/access.2020.3000501.
May 2020, Art. no. 106577, doi: 10.1016/j.compeleceng.2020.106577. [41] J. Yu, R. Li, Z. Feng, A. Zhao, Z. Yu, Z. Ye, and J. Wang, ‘‘A novel
[23] A. Rubaai, M. J. Castro-Sitiriche, and A. R. Ofoli, ‘‘DSP-based laboratory parallel ant colony optimization algorithm for warehouse path plan-
implementation of hybrid fuzzy-PID controller using genetic optimization ning,’’ J. Control Sci. Eng., vol. 2020, Aug. 2020, Art. no. 5287189, doi:
for high-performance motor drives,’’ IEEE Trans. Ind. Appl., vol. 44, no. 6, 10.1155/2020/5287189.
pp. 1977–1986, Nov. 2008, doi: 10.1109/TIA.2008.2006347. [42] M. S. Semary, M. E. Fouda, H. N. Hassan, and A. G. Radwan, ‘‘Realization
[24] I. Pan, S. Das, and A. Gupta, ‘‘Tuning of an optimal fuzzy PID con- of fractional-order capacitor based on passive symmetric network,’’ J. Adv.
troller with stochastic algorithms for networked control systems with Res., vol. 18, pp. 147–159, Jul. 2019, doi: 10.1016/j.jare.2019.02.004.
random time delay,’’ ISA Trans., vol. 50, no. 1, pp. 28–36, Jan. 2011, doi:
[43] G. Tsirimokou, C. Psychalinos, and A. S. Elwakil, ‘‘Emulation of a con-
10.1016/j.isatra.2010.10.005.
stant phase element using operational transconductance amplifiers,’’ Anal.
[25] J. A. Ali, M. A. Hannan, A. Mohamed, and M. G. M. Abdolrasol, ‘‘Fuzzy Integr. Circuits Signal Process., vol. 85, no. 3, pp. 413–423, Dec. 2015,
logic speed controller optimization approach for induction motor drive doi: 10.1007/s10470-015-0626-8.
using backtracking search algorithm,’’ Measurement, vol. 78, pp. 49–62,
[44] M. H. Khooban, M. Shasadeghi, T. Niknam, and F. Blaabjerg, ‘‘Anal-
Jan. 2016, doi: 10.1016/j.measurement.2015.09.038.
ysis, control and design of speed control of electric vehicles delayed
[26] O. Castillo, F. Valdez, J. Soria, L. Amador-Angulo, P. Ochoa, and
model: Multi-objective fuzzy fractional-order PIλ Dµ controller,’’ IET Sci.,
C. Peraza, ‘‘Comparative study in fuzzy controller optimization using bee
Meas. Technol., vol. 11, no. 3, pp. 249–261, May 2017, doi: 10.1049/iet-
colony, differential evolution, and harmony search algorithms,’’ Algo-
smt.2016.0277.
rithms, vol. 12, no. 1, pp. 1–21, Jan. 2019, doi: 10.3390/a12010009.
[45] V. Sharma and S. Purwar, ‘‘Nonlinear controllers for a light-weighted all-
[27] S. Das, I. Pan, S. Das, and A. Gupta, ‘‘A novel fractional order fuzzy
electric vehicle using Chebyshev neural network,’’ Int. J. Veh. Technol.,
PID controller and its optimal time domain tuning based on integral
vol. 2014, Apr. 2014, Art. no. 867209, doi: 10.1155/2014/867209.
performance indices,’’ Eng. Appl. Artif. Intell., vol. 25, no. 2, pp. 430–442,
[46] P. Shah and S. Agashe, ‘‘Review of fractional PID controller,’’
Mar. 2012, doi: 10.1016/j.engappai.2011.10.004.
Mechatronics, vol. 38, pp. 29–41, Sep. 2016, doi: 10.1016/j.
[28] V. Kumar, K. P. S. Rana, and P. Mishra, ‘‘Robust speed control of hybrid
mechatronics.2016.06.005.
electric vehicle using fractional order fuzzy PD and PI controllers in
cascade control loop,’’ J. Franklin Inst., vol. 353, no. 8, pp. 1713–1741, [47] I. Pan and S. Das, ‘‘Kriging based surrogate modeling for fractional order
May 2016, doi: 10.1016/j.jfranklin.2016.02.018. control of microgrids,’’ IEEE Trans. Smart Grid, vol. 6, no. 1, pp. 36–44,
[29] A. Zamani, S. M. Barakati, and S. Yousofi-Darmian, ‘‘Design of a frac- Jan. 2015, doi: 10.1109/TSG.2014.2336771.
tional order PID controller using GBMO algorithm for load–frequency [48] M. S. Mahmoud, Fuzzy Control, Estimation and Diagnosis: Single and
control with governor saturation consideration,’’ ISA Trans., vol. 64, Interconnected Systems. Riyadh, Saudi Arabia: Springer, 2018.
pp. 56–66, Sep. 2016, doi: 10.1016/j.isatra.2016.04.021. [49] H. Bin Duan, D. Bo Wang, and X. Fen Yu, ‘‘Novel approach to nonlinear
[30] M. Al-Dhaifallah, N. Kanagaraj, and K. S. Nisar, ‘‘Fuzzy fractional- PID parameter optimization using ant colony optimization algorithm,’’
order PID controller for fractional model of pneumatic pressure sys- J. Bionic Eng., vol. 3, no. 2, pp. 73–78, Jun. 2006, doi: 10.1016/S1672-
tem,’’ Math. Probl. Eng., vol. 2018, Jan. 2018, Art. no. 5478781, doi: 6529(06)60010-3.
10.1155/2018/5478781. [50] S. Das, I. Pan, and S. Das, ‘‘Performance comparison of optimal fractional
[31] X. Wu, Y. Xu, J. Liu, C. Lv, J. Zhou, and Q. Zhang, ‘‘Characteristics anal- order hybrid fuzzy PID controllers for handling oscillatory fractional
ysis and fuzzy fractional-order PID parameter optimization for primary order processes with dead time,’’ ISA Trans., vol. 52, no. 4, pp. 550–566,
frequency modulation of a pumped storage unit based on a multi-objective Jul. 2013, doi: 10.1016/j.isatra.2013.03.004.
gravitational search algorithm,’’ Energies, vol. 13, no. 1, p. 137, Dec. 2019, [51] S. Kapoulea, C. Psychalinos, and A. S. Elwakil, ‘‘Realizations of
doi: 10.3390/en13010137. simple fractional-order capacitor emulators with electronically-
[32] I. Dimeas, I. Petras, and C. Psychalinos, ‘‘New analog implementa- tunable capacitance,’’ Integration, vol. 69, pp. 225–233, Nov. 2019,
tion technique for fractional-order controller: A DC motor control,’’ doi: 10.1016/j.vlsi.2019.04.004.
AEU-Int. J. Electron. Commun., vol. 78, pp. 192–200, Aug. 2017, doi: [52] S. A. Tekin and M. Alçı, ‘‘Design and applications of electroni-
10.1016/j.aeue.2017.03.010. cally tunable floating resistor using differential amplifier,’’ Electron.
[33] S. Kapoulea, V. Bizonis, P. Bertsias, C. Psychalinos, A. Elwakil, and Electr. Eng., vol. 19, no. 4, pp. 41–46, Apr. 2013, doi: 10.5755/j01.eee.
I. Petráš, ‘‘Reduced active components count electronically adjustable 19.4.1310.
fractional-order controllers: Two design examples,’’ Electronics, vol. 9, [53] Q. Huang, Z. Huang, and H. Zhou, ‘‘Nonlinear optimal and robust speed
no. 1, p. 63, Jan. 2020, doi: 10.3390/electronics9010063. control for a light-weighted all-electric vehicle,’’ IET Control Theory Appl.,
[34] S. Kapoulea, C. Psychalinos, and A. S. Elwakil, ‘‘Single active ele- vol. 3, no. 4, pp. 437–444, Apr. 2009, doi: 10.1049/iet-cta.2007.0367.
ment implementation of fractional-order differentiators and integrators,’’ [54] P. Anantachaisilp and Z. Lin, ‘‘Fractional order PID control of rotor
AEU-Int. J. Electron. Commun., vol. 97, pp. 6–15, Dec. 2018, doi: suspension by active magnetic bearings,’’ Actuators, vol. 6, no. 1, pp. 1–31,
10.1016/j.aeue.2018.09.046. Jan. 2017, doi: 10.3390/act6010004.

VOLUME 9, 2021 73411

 M. A. George et al.: Electronically Tunable ACO Based Fuzzy FOPID Controller for Effective Speed Control of EV

[55] A. X. R. Irudayaraj, N. I. A. Wahab, M. G. Umamaheswari, DATTAGURU V. KAMAT (Senior Member, IEEE)

M. A. M. Radzi, N. B. Sulaiman, V. Veerasamy, S. C. Prasanna, and was born in India, in 1965. He received the B.E.
R. Ramachandran, ‘‘A Matignon’s theorem based stability analysis degree in electronics and communication and the
of hybrid power system for automatic load frequency control using M.E. degree in digital electronics from the B.V.
atom search optimized FOPID controller,’’ IEEE Access, vol. 8, Bhoomaraddi College of Engineering, Karnataka
pp. 168751–168772, 2020, doi: 10.1109/ACCESS.2020.3021212. University, Dharwad, in 1987 and 1997, respec-
[56] D. Qian, C. Li, R. P. Agarwal, and P. J. Y. Wong, ‘‘Stability analy- tively, and the Ph.D. degree in analogue VLSI
sis of fractional differential system with Riemann–Liouville derivative,’’
signal processing from the Manipal Academy
Math. Comput. Model., vol. 52, nos. 5–6, pp. 862–874, Sep. 2010, doi:
of Higher Education (MAHE), Manipal, India,
10.1016/j.mcm.2010.05.016.
[57] G. Tsirimokou, A. Kartci, J. Koton, N. Herencsar, and C. Psychalinos, in 2013.
‘‘Comparative study of discrete component realizations of fractional- He is currently working as a Professor with the Department of Electronics
order capacitor and inductor active emulators,’’ J. Circuits, Syst. and Communication Engineering, Manipal Institute of Technology, Manipal.
Comput., vol. 27, no. 11, Oct. 2018, Art. no. 1850170, doi: His research contribution includes one U.S. patent granted at WIPO, more
10.1142/S0218126618501700. than 20 full regular articles published in indexed international journals,
and 20 indexed international conference publications. His research interests
include digital, analog, and mixed-signal VLSI design, analog VLSI signal
processing, digital VLSI architectures, and fractional-order circuits.
Dr. Kamat is a member of IET, IE (India), and ISTE (India).

CIJI PEARL KURIAN (Senior Member, IEEE)

was born in India, in 1964. She received the
B.Tech. degree in electrical and electronics
engineering from Calicut University, in 1986,
the M.Tech. degree in lighting science and engi-
MARY ANN GEORGE (Graduate Student Mem- neering from Mangalore University, in 1994, and
ber, IEEE) was born in India, in 1991. She received the Ph.D. degree in electrical engineering from
the bachelor’s degree in electronics and communi- Manipal University, Manipal, India, in 2007.
cation engineering (ECE) and the master’s degree Since 1987, she has been teaching with the
in digital electronics and advanced communication Department of Electrical and Electronics Engi-
from the Manipal Institute of Technology (MIT), neering, Manipal Institute of Technology, Manipal, a constituent institution
Manipal Academy of Higher Education (MAHE), of the Manipal Academy of Higher Education, India. Her research interest
India, in 2013 and 2016, respectively. She is cur- includes lighting controls technology and applications.
rently pursuing the Ph.D. degree with the ECE Dr. Kurian is a Fellow of the Institution of Engineers, India, and a Life
Department, MIT, Manipal. Member of professional bodies, including the Indian Society of Lighting
Her research interests include fractional-order controllers, fractional-order Engineers, the Indian Society for Technical Education, and the Systems
systems, and analog circuits. She is a Life Member of the Indian Society of Society of India.
Systems for Science and Engineering (ISSE).

73412 VOLUME 9, 2021