Journal of Energy Storage 42 (2021) 103079

Contents lists available at ScienceDirect

Journal of Energy Storage

journal homepage: www.elsevier.com/locate/est

Robust integral backstepping controller for energy management in plugin

hybrid electric vehicles
Zil e Huma a , Muhammad Kashif Azeem a , Iftikhar Ahmad a ,∗, Hammad Armghan b ,
Shahzad Ahmed a , Hafiz Mian Muhammad Adil a
a
School of Electrical Engineering and Computer Science (SEECS), National University of Sciences and Technology (NUST), Islamabad, Pakistan
b School of Electrical Engineering, Shandong University, Jinan, 250061, China

ARTICLE INFO ABSTRACT

Keywords: Plugin hybrid electric vehicles (PHEVs) are a suitable choice to achieve enhanced performance and reduced
Plugin hybrid electric vehicles toxic gases. The considered PHEV consists of an integrated charging unit and a hybrid energy storage system
Hybrid energy storage system (HESS). The proposed HESS comprises of a battery with high energy density and a supercapacitor with high
Battery
power density coupled together to fulfill the load demands of the vehicle. A DC–DC buck converter with an
Supercapacitor
uncontrolled rectifier has been used for an on-board and balanced charging and two bi-directional DC–DC
Robust integral backstepping
Supervisory control
buck-boost converters have been incorporated to ensure a smooth transition of energy. A rule-based algorithm
DC–DC converters has been employed as a supervisory controller by incorporating total power inflow and state of charge of the
State of charge power sources to fulfill the load demands. Moreover, a robust integral backstepping-based nonlinear controller
has been designed for the smooth execution and energy management of PHEV in terms of output voltage
regulation, reference generation, and smooth tracking of current. Lyapunov stability theory has been used to
ensure asymptotic stability of the PHEV. The performance of the proposed controller has been validated using
simulations on MATLAB/Simulink by comparing it with the Lyapunov redesign and backstepping controllers
already proposed in the literature. The robustness of the proposed controller has been verified by introducing
uncertainty in the state model. Finally, the real-time applicability and effectiveness of the proposed work have
been ensured using controller hardware in loop (C-HIL) test bench.

1. Introduction converter will be able to function as an AC/DC battery charger with a

reduced number of high-current inductors and current transducers.
The issues of regular increase in the concentration of CO2 in the Generally, two types of storage elements are used in the HEVs: one
environment and the rapid increase in global warming are the major with higher energy storing capacity (primary source) and the second
issues observed in the past few decades, needed to be addressed [1–3]. with enhanced power delivering ability (secondary source) [10]. PHEVs
Transportation vehicles have been observed to be the major contrib- with battery as an additional power source along with an internal com-
utors to the greenhouse gases (GHG). To retain a clean environment bustion engine (ICE) is conventionally being used [11]. The presence of
by using evergreen energy resources and efficient utilization of power ICEs still contribute enormously to the environmental pollution. Hence,
sources, along with hybrid energy storage systems are the major con- there is a need for the integration of auxiliary power source with the
cerns of today’s researchers [4,5]. In these circumstances, PHEVs are battery and the removal of ICEs.
found to be the better alternative as compared to conventional vehicles
Several combinations of the hybrid HESS for efficient energy man-
because of reduced pollution and efficient performance [6]. The PHEVs
agement have been presented in the literature [12–15]. These systems
are the replica of hybrid electric vehicles (HEVs) having an additional
hold benefits of enhanced energy storing capacity, stabilized perfor-
ability to charge from an external charging unit [7,8]. On the other
mance, smart energy utilization through regenerative braking, and
hand, the use of battery chargers with an onboard arrangement would
smooth operating capability with dynamic load profile. For static load
allow battery charging at any time given the availability of the supply
profile, the proton exchange membrane fuel cell (PEMFC) is used
grid. A novel integrated bidirectional AC/DC charger for PHEVs is an
essence to reduce the cost and increased reliability [9]. The integrated which is inefficient when operated under power transients [16,17].

∗ Corresponding author.
E-mail addresses: [email protected] (Z.e. Huma), [email protected] (M.K. Azeem), [email protected] (I. Ahmad),
[email protected] (H. Armghan), [email protected] (S. Ahmed), [email protected] (H.M.M. Adil).

https://doi.org/10.1016/j.est.2021.103079
Received 18 January 2021; Received in revised form 6 July 2021; Accepted 9 August 2021
Available online 20 August 2021
2352-152X/© 2021 Elsevier Ltd. All rights reserved.
Z.e. Huma et al. Journal of Energy Storage 42 (2021) 103079

Furthermore, for the dynamic load profile, fuel cell (FC) along with converter formulating the integrated charging unit which is linked to
ultracapacitor (UC) has been used to overcome the transients [18,19]. a DC–DC converter and a hybrid energy storage system comprising of
Moreover, the combination of FC–battery–UC helps in increased storage a battery and supercapacitor was directly linked to the DC bus with
capacity and enhanced performance with truncated load demands [20, two DC–DC converters. Here 𝑃𝑏𝑎𝑡 , 𝑃𝑠𝑐 , 𝑉𝑏𝑎𝑡 , 𝑉𝑠𝑐 , 𝐼𝑏𝑎𝑡 , 𝐼𝑠𝑐 and 𝑆𝑜𝐶𝑏𝑎𝑡 ,
21]. But in general, this contributes enormously to the computational 𝑆𝑜𝐶𝑠𝑐 represent the power, voltage, current and state of charge of the
complexity and cost of the system. battery and ultracapacitor, respectively whereas 𝑢23 and 𝑢45 denote
To ensure stable operation, the battery is commonly used in electric the nonlinear control signals. Furthermore, the state variables, and
vehicles as a single storage device, but it lacks in contributing towards errors are shown by 𝑋𝑖 and 𝑒𝑖 , respectively. The reason for using the
variations in the load profile. We can optimize batteries for high uncontrolled boost rectifier is to allow the power flow in both the
energy capacity and consequently slightly reduced power capacity but a directions, from the grid to the vehicle and vehicle to the grid along
combination of battery and supercapacitor (SC) can also be used. Hence with the AC–DC conversion. Moreover, a buck converter is bridged to
a combination of battery with the SC has been used to overcome all the the boost rectifier circuit in order to ensure the regulated charging
stated issues. In this arrangement, the battery will be used for enhanced voltages across the charger. Furthermore, a HESS has been utilized
storage and SC will cater for the power transients with its high-power considering the proposed energy management algorithm to extend the
delivering capability [22,23]. driving range and life of the storage elements. The HESS consisting of
Various controllers for the energy management of HEVs have been battery, SC bank, and DC–DC buck-boost converters is further linked
presented in the literature. The dynamic behavior of the storage el- to the DC line to power up the motors of the vehicle. Depending upon
ements has been observed and energy management has been done the performance characteristics, the control algorithm is split into two
using an artificial neural network (ANN) with a PI controller in parallel portions:
to capture the regenerative current [24]. A multi-input–multi-output
(MIMO) bidirectional converter with a PI controller for the reference • Nonlinear Control: This control part guarantees the output volt-
tracking of current and voltage of the power sources is also consid- age regulation, asymptotic stability, and current tracking of the
ered [25]. The fractional-order PI controller [26] has been used to cater energy storage elements to their respective reference values for
for the uncertainties present in the HESS. Due to the irregular switching the energy management of PHEV.
behavior of DC–DC converters, the mathematical models derived for • Supervisory Control: In this portion, the algorithm is defined
HEVs are actually nonlinear which are linearized around a specific depending upon the load requirements, and is done by balancing
equilibrium point in most of these works, hence limiting the flexibility the power at the load and source sides. The SoC of the energy
of operation [27]. storage unit is used to capture the excessive power.
Droop control is the commonly used method for the smooth voltage
regulation, [28] which has its limitation for the simultaneous operation The major contributions of this research paper are:
of multiple storage units with a different state of charge (SoC) [29]. (1) Voltage regulation of the charger and the DC bus with varying
A nonlinear model-based fuzzy sliding mode control has been imple- load dynamics of the PHEV.
mented and is used to overcome the short-range issue of HEV [30]. (2) Absolute reference generation of the current along with reg-
Two 𝐻∞ controllers have been proposed for the patterned charging and ulated charging and discharging profiles for the primary and
discharging of the supercapacitor [31]. The detailed stability analysis secondary sources.
has not been carried out in most of these formulated systems. The (3) Asymptotic stability of the PHEV under various operating cir-
adaptive model predictive control (AMPC) [32] has been utilized for cumstances.
predictive energy management of the HESS but the integrated charging (4) Estimation of the gains using a genetic algorithm for the optimal
unit has not been taken into account. Nonlinear controller along with approximation of the controller gains.
adaptation law has been used for the multi-source HESS but the defined (5) Ensure the robust and instantaneous response of the proposed
sliding surfaces were not assumed to be adaptive [21]. Numerous con- controller in the presence of model uncertainties.
trol strategies have been proposed in the literature for the PHEVs but (6) Verification of the proposed controller using MATLAB/Simulink
are confined to the isolated mode, hence not considering the rule-based and real-time controller hardware in the loop setup.
strategy along with the nonlinear controllers [33–37].
The modeling and control of charger with the real-time equivalent This research article is organized as follows. Section 2 provides
model of the battery and HESS on a single unit, considering the the mathematical state model of the integrated charger along with
irregularly varying load profile has always been a challenge. Moreover, the HESS for PHEV. The design procedure for the proposed RIBS
the stability of these systems has always been a tough job to deliver. controller has been elaborated in Section 3. In Section 4, a rule-based
Hence, a more versatile and robust nonlinear controller in accordance supervisory controller for the energy management of PHEV has also
to the supervisory controller is required for the PHEV to operate been defined. Section 5 summarizes the detailed simulation results and
within the specified limit of operation. Few standardized nonlinear their analyses obtained through MATLAB/Simulink and comparison
controllers have been implemented to validate the performance of the with multiple controllers proposed in the literature. In Section 6, a
proposed controller [10,20]. Control Lyapunov function (CLF) does not real-time controller hardware in loop analysis has been done for the
behave in the prescribed limits of the system resulting in unnecessary verification of the proposed controller. Finally, Section 7 concludes the
overshoots/undershoots during the load transition phase. Furthermore, research article.
the backstepping control technique being a systematic recursive ap-
proach for the reference tracking has been used. This results in reduced 2. Mathematical representation of PHEV
amplitude for the transients during load variations, but an irregular
response in terms of oscillations around the reference point and some 2.1. Design environment of integrated charging unit
steady-state error is observed.
In this research article robust integral backstepping (RIBS) based With the growing demand for PHEVs, it is necessary to develop an
nonlinear controller along with a rule-based strategy for the supervisory integrated charging source to charge the vehicle. The general architec-
controller, has been proposed to overcome all the stated issues. The ture of the smart charger using a unidirectional DC–DC converter is
asymptotic stability of the considered model has been affirmed using shown in Fig. 2. Various charging topologies for PHEVs are available
the Lyapunov stability theory. The general model of PHEV depicted in in the literature [38,39]. The integration of the charger along with the
Fig. 1 consists of an uncontrolled boost rectifier, bridged to the buck energy storage unit leads to the optimized performance with reduced

2
Z.e. Huma et al. Journal of Energy Storage 42 (2021) 103079

Fig. 1. General architecture of PHEV.

cost. Moreover, this arrangement places a great impact by targeting

the power distribution system in terms of power factor correction and
harmonic oscillations [40]. Therefore, a novel idea of an integrated
charging unit has been used. The output of the charger is described
in terms of charging voltages 𝑉𝑐ℎ and the current 𝐼𝑐ℎ . The DC–DC buck
converter in use is attached directly to the battery, being the only direct
charging source. The obtained voltages are definite function of SoC and
is mathematically defined as:
𝐶ℎ
𝑆𝑜𝐶 = (1)
𝐶ℎ𝑛𝑜𝑚
where SoC is the state of charge with 𝐶ℎ and 𝐶ℎ𝑛𝑜𝑚 as an absolute Fig. 2. Electric schema of AC–DC power charger for the PHEV.
and nominal capacity in ‘‘Ah’’ of the battery respectively. The termi-
nal voltage of the battery can be determined by using the dynamic
parameters, and it depends on the battery SoC and impedance. Also, the
be operating in continuous conduction mode (CCM). The components
energy management algorithm helps in estimating the battery current
convoyed by the uncontrolled boost rectifier comprise of an AC–DC
by monitoring the SoC of the battery which can be estimated using
rectifier, a diode 𝐷𝑝𝑓 𝑐 , a switch 𝑆𝑝𝑓 𝑐 , an inductor 𝐿𝑝𝑓 𝑐 , and a capacitor
this equation. The active and reactive power of the supply during 𝐶𝑝𝑓 𝑐 to capture the ripples. Furthermore, the controlled switching
consecutive cycles of power equilibrium is expressed as [41]: converter constitutes of the basic electronic components: a nonlinear
𝑃𝑔𝑟𝑖𝑑 = 𝑉 𝐼𝑐𝑜𝑠𝜙 (2) control switch 𝑆1 , output filtering capacitor 𝐶1 and an inductor 𝐿1 . The
equivalent circuit of the battery consists of equivalent series resistance
𝑃𝑔𝑟𝑖𝑑 = 𝑉 𝐼𝑠𝑖𝑛𝜙 (3) 𝑅𝐵 , equivalent parallel resistance 𝑅𝑃 and the capacitance of the battery
𝐶𝐵 . As the switching behavior of the converter is nonlinear, hence the
where 𝑃𝑔𝑟𝑖𝑑 is the power from the grid, 𝜙 being the phase angle reflects
nonlinear controller must be designed. The system under consideration
the phase shift between grid voltage 𝑉 and the current 𝐼. The efficiency
is modeled using inductor volt second and capacitor charge balance
of the charger in terms of DC voltage and active AC power is considered theorems and is mathematically represented in terms of desired state
as: dynamical model:
𝑃𝑐ℎ = 𝜂[𝑉 𝐼𝑐𝑜𝑠𝜙] (4) 𝑑𝑥1 𝑉𝑔 𝑥
= 𝑢 − 2 (5)
𝑑𝑡 𝐿1 1 𝐿1
where the power from the charger is captured as 𝑃𝑐ℎ with 𝜂 being the ( )
efficiency of the charging source. 𝑑𝑥2 𝑥 𝑥2 − 𝑥3
= 1 − (6)
The integrated charging unit consists of the mains supply concate- 𝑑𝑡 𝐶1 𝑅 𝐵 𝐶1
( )
nated with an uncontrolled boost rectifier and a DC–DC buck converter. 𝑑𝑥3 𝑥2 1 1
= − + 𝑥3 (7)
To approximate the charging and discharging characteristics of the 𝑑𝑡 𝑅 𝐵 𝐶𝐵 𝑅 𝐵 𝐶𝐵 𝑅 𝑃 𝐶𝐵
battery, its equivalent model is considered as shown in Fig. 2. The smart Here 𝑥1 , 𝑥2 and 𝑥3 are average inductor current 𝐼𝐿1 , the average
charger supplies the deficit power to the battery capsuled in an energy output voltage of charger 𝑉𝑐ℎ and the battery inner voltage 𝑉𝐵 respec-
management system (EMS), as directed by the supervisory control. The tively. Also 𝑉𝑔 are the voltage from the AC plug and 𝑢1 is the nonlinear
controlled buck converter used in the charger has been assumed to PWM signal.

3
Z.e. Huma et al. Journal of Energy Storage 42 (2021) 103079

𝑑𝐼𝑏𝑎𝑡 𝑉 𝑅 𝑉
= 𝑏𝑎𝑡 − 2 𝐼𝑏𝑎𝑡 − 𝑢23 𝑑𝑐 (14)
𝑑𝑡 𝐿2 𝐿2 𝐿2
The state model represented by Eqs. (9)–(12) can hence be written in
the more simplified form by incorporating Eq. (13) as follows:
𝑑𝑉𝑑𝑐 𝐼 𝐼
= 𝑢23 𝑏𝑎𝑡 − 1 (15)
𝑑𝑡 𝐶𝑑𝑐 𝐶𝑑𝑐
SC is used to supply power during the load transients and is
charged through regenerative braking and excessive power from the
main source. Being a high power density device, it cannot be charged
directly through the mains. This bidirectional process of charging and
discharging will be accomplished using the similar DC–DC converter
as of battery. The similarity between both the converters is of quite
importance and is used to define the state dynamical model. Hence the
switching factor 𝑁 can be defined as:
Fig. 3. Schematic diagram of HESS.
{
1, if (𝐼𝑠𝑐𝑟𝑒𝑓 > 0)
𝑁= (16)
0, if (𝐼𝑠𝑐𝑟𝑒𝑓 < 0)
2.2. State representation of HESS
where 𝐼𝑠𝑐𝑟𝑒𝑓 is the reference current of the SC. The contribution of the
current from auxiliary source (𝐼𝑠𝑐 ) can be represented as:
The general architecture for HESS under consideration has been
shown in Fig. 3 which comprises of two power-up sources, labeled 𝐼2 = (1 − 𝑢4 )𝐼𝑠𝑐 (17)
as the primary and secondary sources. Moreover, these sources are in
direct contact with the DC–DC converters. Two bi-directional buck- where 𝑢4 denotes the control effort. The fact for the charging (regenera-
boost converters are attached to the common DC bus, proceeding to the tive braking) 𝐼𝑠𝑐𝑟𝑒𝑓 < 0 and discharging 𝐼𝑠𝑐𝑟𝑒𝑓 > 0 of SC can be captured
predefined load. The DC–DC converter attached to the battery consists using following state equations of SC current:
of the basic electronic components: A common filtering capacitor 𝐶𝑜 𝑑𝐼𝑠𝑐 𝑉 𝑅 𝑉
coupled directly to the DC bus, two PWM controlled switches labeled = 𝑠𝑐 − 3 𝐼𝑠𝑐 − (1 − 𝑢4 ) 𝑑𝑐 (18)
𝑑𝑡 𝐿3 𝐿3 𝐿3
as 𝑆2 and 𝑆3 and an inductor 𝐿2 with an internal resistance 𝑅2 . To 𝑑𝐼𝑠𝑐 𝑉𝑠𝑐 𝑅3 𝑉𝑑𝑐
attain the regulated voltage at the DC bus, pulses from the nonlinear = − 𝐼 − 𝑢5 (19)
𝑑𝑡 𝐿3 𝐿3 𝑠𝑐 𝐿3
controller are provided to the gate of the semiconductor switches.
To describe the charging and discharging behavior of the battery 𝐼2 = 𝑢5 𝐼𝑠𝑐 (20)
during varying load demands, the following mathematical viability is where 𝑢5 describes the nonlinear control signal. To obtain the global
ensured: model of the system, dual control 𝑢45 for the converter is defined as:
{
1, if (𝐼𝑏𝑎𝑡𝑟𝑒𝑓 > 0)
𝑀= (8) 𝑢45 = [𝑁(1 − 𝑢4 ) + (1 − 𝑁)𝑢5 ] (21)
0, if (𝐼𝑏𝑎𝑡𝑟𝑒𝑓 < 0)
The mutated state dynamics of the system is stated as:
where the term 𝐼𝑏𝑎𝑡𝑟𝑒𝑓 is used to describe the desired current of the
𝑑𝐼𝑠𝑐 𝑉 𝑅 𝑉
battery which will further be utilized to estimate the present power = 𝑢𝑐 − 3 𝐼𝑢𝑐 − 𝑢45 𝑑𝑐 (22)
deliverance capability of the source using the initial state of charge 𝑑𝑡 𝐿3 𝐿3 𝐿3
(SoC). 𝐼2 = 𝑢45 𝐼𝑠𝑐 (23)
As the DC–DC converter employed is intrinsically bi-directional,
hence its dual operation is specified by the maxima and minima of the Furthermore, to access the cumulative current at the DC line Kirchhoff’s
load profile. When the control signal is applied at switch 𝑆2 , shifting it current rule is applied at the common node.
to ON condition with its counter 𝑆3 to operate in OFF state, The battery 𝐼𝑜 = 𝐼1 + 𝐼2 (24)
will tend to perform in boost mode (𝐼𝑏𝑎𝑡𝑟𝑒𝑓 > 0) and for (𝐼𝑏𝑎𝑡𝑟𝑒𝑓 < 0),
the battery will conduct in buck mode at an inverse rate to maintain By placing the value of 𝐼2 from Eq. (23) in Eq. (24) will give
the power equilibrium.
𝐼1 = 𝐼𝑜 − 𝑢45 𝐼𝑠𝑐 (25)
The averaged state dynamical model of the battery, associated
with buck-boost converter during charging and discharging modes has Now, to obtain the differential equation for DC bus voltage 𝑉𝑑𝑐 , putting
been defined using inductor volt second and capacitor charge balance value of 𝐼1 from Eq. (25) in Eq. (15), we have
theorems as: 𝑑𝑉𝑑𝑐 𝐼 𝐼 𝐼
𝑑𝐼𝑏𝑎𝑡 𝑉 𝑅 𝑉 = 𝑢23 𝑏𝑎𝑡 + 𝑢45 𝑠𝑐 − 𝑜 (26)
= 𝑏𝑎𝑡 − 2 𝐼𝑏𝑎𝑡 − (1 − 𝑢2 ) 𝑑𝑐 (9) 𝑑𝑡 𝐶𝑑𝑐 𝐶𝑑𝑐 𝐶𝑑𝑐
𝑑𝑡 𝐿2 𝐿2 𝐿2
It is worthwhile to mention that the perceptible states 𝑥4 , 𝑥5 and 𝑥6
𝑑𝑉𝑑𝑐 𝐼 𝐼
= (1 − 𝑢2 ) 𝑏𝑎𝑡 − 1 (10) are defined as 𝐼𝑏𝑎𝑡 , 𝐼𝑠𝑐 and 𝑉𝑑𝑐 . Where 𝑢23 and 𝑢45 are nonlinear control
𝑑𝑡 𝐶𝑑𝑐 𝐶𝑑𝑐 inputs. By using the Eqs. (14), (22) and (26), the globalized state model
𝑑𝐼𝑏𝑎𝑡 𝑉 𝑅 𝑉
= 𝑏𝑎𝑡 − 2 𝐼𝑏𝑎𝑡 − 𝑢3 𝑑𝑐 (11) of HESS can be articulated as:
𝑑𝑡 𝐿2 𝐿2 𝐿2 𝑑𝑥4 𝑉 𝑅 𝑥
𝑑𝑉𝑑𝑐 𝐼 𝐼 = 𝑏𝑎𝑡 − 2 𝑥4 − 𝑢23 6 (27)
= 𝑢3 𝑏𝑎𝑡 − 1 (12) 𝑑𝑡 𝐿2 𝐿2 𝐿2
𝑑𝑡 𝐶𝑑𝑐 𝐶𝑑𝑐 𝑑𝑥5 𝑉 𝑅 𝑥
where 𝐼𝑏𝑎𝑡 and 𝐼1 represent battery current and output current re- = 𝑢𝑐 − 3 𝑥5 − 𝑢45 6 (28)
𝑑𝑡 𝐿3 𝐿3 𝐿3
spectively. Whereas 𝑉𝑏𝑎𝑡 , and 𝑉𝑑𝑐 are battery and DC bus voltage 𝑑𝑥6 𝑥 𝑥 𝐼
respectively. Also, 𝑢2 and 𝑢3 are the nonlinear control inputs. = 𝑢23 4 + 𝑢45 5 − 𝑜 (29)
𝑑𝑡 𝐶𝑑𝑐 𝐶𝑑𝑐 𝐶𝑑𝑐
Furthermore, the compact form of the designated model can be
The state equations of multi-input–multi-output (MIMO) systems are
obtained by defining the virtual control 𝑢23 as:
dynamically nonlinear and hence design of a nonlinear controller is
𝑢23 = [𝑀(1 − 𝑢2 ) + (1 − 𝑀)𝑢3 ] (13) necessary.

4
Z.e. Huma et al. Journal of Energy Storage 42 (2021) 103079

3. Nonlinear robust integral backstepping based controller Placing the value of 𝑠̇ 1 from Eq. (39) in Eq. (34) produces:
𝑠 𝑠
𝑉̇ 1 = −𝑘1 𝑠1 𝑠𝑔𝑛(𝑠1 ) + 1 2 (40)
3.1. Controller design of DC-DC converters for integrated charger 𝐶1
By taking the time derivative of Eq. (38), we get:
The converters associated with the charging unit should be ade-
quately controlled to ensure voltage regulation at the intermediate DC 𝑠̇ 2 = 𝑥̇ 1 − 𝛼̇ (41)
bus. The proposed robust integral backstepping controller maintains In the expression of 𝛼 defined by Eq. (35), the signum function can be
the ability of both the integral backstepping and sliding mode control replaced by saturation function to avoid chattering, because the value
(SMC). Integral backstepping is a recursive approach for minimization of sat function varies uniformly around the mean point and is defined
of error in steady-state. Whereas, SMC is used for the reduction of as [42]:
convergence time of a signal and robustness against transients in terms 𝑠1
of load variations and bounded uncertainties. To ensure the efficient 𝑠𝑔𝑛(𝑠1 ) = (42)
|𝑠1 | + 𝜇1
tracking of the charging voltage for the battery, the sliding surface 𝑠1
Here 𝜇1 defines the isolated operating region of the charger. Now
in terms of state variable 𝑥2 and its reference is defined as: equation for 𝛼 is rewritten as:
( ( ) )
𝑠1 = 𝑥2 − 𝑥2𝑟𝑒𝑓 (30) 𝑠1 𝑥2 − 𝑥3
𝛼 = −𝑘1 + + 𝑥̇ 2𝑟𝑒𝑓 − 𝛽1 𝜓1 𝐶1 (43)
By taking the time derivative of 𝑠1 defined in Eq. (30) and substi- |𝑠1 | + 𝜇1 𝑅 𝐵 𝐶1
tuting value of 𝑥̇ 2 from Eq. (6) yields: Furthermore, differentiating Eq. (43) with respect to time gives:
( )
𝑥 𝑥2 − 𝑥3 (|𝑠1 | + 𝜇1 )𝑠̇ 1 − 𝑠1 (|𝑠̇ 1 |) 𝑉𝑔 𝑥2
𝑠̇ 1 = 1 − − 𝑥̇ 2𝑟𝑒𝑓 (31) 𝛼̇ = − 𝑘1 𝐶1 + 𝑢̇ 1 −
𝐶1 𝑅 𝐵 𝐶1 (|𝑠1 | + 𝜇1 )2 𝑅𝐵 𝑅2𝐵 𝐶𝐵
( )
Moreover, to reduce the chattering the sliding surface with an integral 1 1
+ + 𝑥3 + 𝑥̈ 2𝑟𝑒𝑓 𝐶1 − 𝛽1 𝜓̇ 1 𝐶1 (44)
action is defined as: 𝑅2𝐵 𝐶𝐵 𝑅 𝐵 𝑅 𝑃 𝐶𝐵
1
where 𝑚 is defined for the ease of calculation and can be written as:
𝜓1 = (𝑥2 − 𝑥2𝑟𝑒𝑓 )𝑑𝑡 (32)
∫0 (|𝑠 | + 𝜇1 )𝑠̇ 1 − 𝑠1 (|𝑠̇ 1 |) 𝑥2
𝑚 =𝑘1 𝐶1 1 +
𝜓̇ 1 = 𝑠1 = 𝑥2 − 𝑥2𝑟𝑒𝑓 (32a) (|𝑠1 | + 𝜇1 )2 𝑅2𝐵 𝐶1
( )
1 1
Now, the Lyapunov candidate function is defined to clinch the stability − + 𝑥3 (45)
of the integrated charging unit as: 𝑅2𝐵 𝐶𝐵 𝑅 𝐵 𝑅 𝑃 𝐶𝐵

1 2 𝛽1 2 By incorporating the values of 𝑥̇ 1 and 𝛼̇ from Eqs. (5) and (44) in

𝑉1 = 𝑠 + 𝜓1 (33) Eq. (41) yields:
2 1 2
where 𝛽1 is the gain of the controller with its value necessarily > 0. 𝑉𝑔 𝑥2 𝑉𝑔
𝑠̇ 2 = 𝑢1 − +𝑚− 𝑢̇ − 𝑥̈ 2𝑟𝑒𝑓 𝐶1 + 𝛽1 𝜓̇ 1 𝐶1 (46)
Now, taking time derivative of 𝑉1 defined in Eq. (33) and substituting 𝐿1 𝐿1 𝑅𝐵 1
the value of 𝑠̇ 1 from Eq. (31) and 𝜓̇ from Eq. (32a) results as follows: Now, the composite Lyapunov function is defined as:
𝑉̇ 1 =𝑠1 𝑠̇ 1 + 𝛽1 𝜓1 𝜓̇ 1 1 2
𝑉2 = 𝑉1 + 𝑠 (47)
( ( ) ) 2 2
𝑥1 𝑥2 − 𝑥3
=𝑠1 − − 𝑥̇ 2𝑟𝑒𝑓 + 𝛽1 𝜓1 (34) By taking the time derivative of Eq. (47a) gives:
𝐶1 𝑅 𝐵 𝐶1
𝑉̇ 2 = 𝑉̇ 1 + 𝑠2 𝑠̇ 2 (47a)
Moreover, for the system to be asymptotically stable, the derivative
of 𝑉1 must be negative semi-definite i.e. 𝑉̇ 1 ≤ 0. Where 𝑥1 will act as Now substituting the values of 𝑉̇ 1 from Eq. (40) and 𝑠̇ 2 from Eq. (46)
virtual control 𝛼 given as: yields:
(
( ) 𝑠 𝑠 𝑉𝑔 𝑥 𝑉𝑔
𝑥2 − 𝑥3 𝑉̇2 = − 𝑘1 |𝑠1 | + 1 2 + 𝑠2 𝑢 − 2 +𝑚− 𝑢̇
𝛼 = (−𝑘1 𝑠𝑔𝑛(𝑠1 ) + + 𝑥̇ 2𝑟𝑒𝑓 − 𝛽1 𝜓1 )𝐶1 (35) 𝐶1 𝐿1 1 𝐿1 𝑅𝐵 1
𝑅𝐵 𝐶1 )
− 𝑥̈ 2𝑟𝑒𝑓 𝐶1 + 𝛽1 𝜓̇ 1 𝐶1 (48)

𝑉̇ 1 = −𝑘1 𝑠1 𝑠𝑔𝑛(𝑠1 ) (36) Further simplification results in:

(
where 𝑘1 is the switching gain of the controller with its value > 1 and 𝑠1 𝑉𝑔 𝑥 𝑉𝑔
𝑉̇2 = − 𝑘1 |𝑠1 |𝑥 + 𝑠2 + 𝑢 − 2 +𝑚− 𝑢̇
signum function is defined as [42]: 𝐶1 𝐿 1 1 𝐿 1 𝑅𝐵 1
{𝑥 )
|𝑥|
, 𝑥≠0 − 𝑥̈ 2𝑟𝑒𝑓 𝐶1 + 𝛽1 𝜓̇ 1 𝐶1 (48a)
𝑠𝑔𝑛(𝑥) = (37)
0, 𝑥=0
To prove 𝑉̇ 2 ≤ 0 and assure the asymptotic stability of the charging
The second error in terms of sliding surface 𝑠2 is defined as: system, following is considered:
𝑠2 = 𝑥 1 − 𝛼 (38) 𝑉̇2 = −𝑘1 |𝑠1 | − 𝑘2 |𝑠2 | (49)
The control objective will be achieved only if the term 𝑠2 converges By solving Eqs. (48) and (49) simultaneously, the control law 𝑢1 is
to zero, which will result in convergence of 𝑠1 , hence, both the errors obtained as:
(
vanish. This will result in a regulated intermediate DC bus for the 𝑅 𝑠2 𝑠 𝑉𝑔 𝑥
𝑢̇ 1 = 𝐵 𝑘2 + 1 + 𝑢 − 2 +𝑚
charger. Putting the value of 𝑥1 from Eq. (38) in Eq. (31) gives: 𝑉𝑔 |𝑠2 | + 𝜇2 𝐶1 𝐿1 1 𝐿1
)
( ) − 𝑥̈ 2𝑟𝑒𝑓 𝐶1 + 𝛽1 𝜓̇ 1 𝐶1 (50)
𝑠2 + 𝛼 𝑥2 − 𝑥3
𝑠̇ 1 = − − 𝑥̇ 2𝑟𝑒𝑓
𝐶1 𝑅 𝐵 𝐶1 where 0 ≤ 𝑢1 ≤ 1. The control law presented in Eq. (50) is used
𝑠2 to generate the duty cycle for the charger. The obtained control law
= − 𝑘1 𝑠𝑔𝑛(𝑠1) + − 𝛽1 𝜓1 (39)
𝐶1 regulates the voltage of the integrated charging unit.

5
Z.e. Huma et al. Journal of Energy Storage 42 (2021) 103079

3.2. Nonlinear controller design of DC-DC converters for HESS To describe the dynamics of the sliding surface, taking time derivative
of Eq. (61) produces:
[ ]
In this subsection, a robust integral backstepping controller is pre- 𝑉 𝑉 𝑥
sented for the energy management of HESS and to achieve the design 𝑠̇ 3 = 𝑏𝑎𝑡 𝑥4 + 𝑢𝑐 𝑥5 − 6 𝐼𝑜 + 𝛾 − 𝑥̇ 6𝑟𝑒𝑓 (62)
𝑥6 𝐶𝑑𝑐 𝑉𝑏𝑎𝑡 𝑉𝑏𝑎𝑡
objectives. The coupled controllers come handy when benefits for
Moreover, the integral action is necessary to reduce the chattering.
dual control techniques are required. For this purpose the state model
Defining the integral term for reference generation of battery current
defined by Eqs. (27–29) for HESS needs to be altered by the variations
and hence differentiating with respect to time results in the following
in terms of uncertainties 𝜃23 and 𝜃45 defined sequentially as:
expressions:
𝑉𝑏𝑎𝑡 𝑥
𝜃23 = − 𝑢23 6 (51) 1
𝐿2 𝐿2 𝜓2 = (𝑥6 − 𝑥6𝑟𝑒𝑓 )𝑑𝑡 (63)
∫0
𝑉 𝑥
𝜃45 = 𝑢𝑐 − 𝑢45 6 (52) 𝜓̇ 2 = 𝑠3 = 𝑥6 − 𝑥6𝑟𝑒𝑓 (63a)
𝐿3 𝐿3
By performing some algebraic simplifications, the dual control defined Now the composite Lyapunov candidate function is defined to ensure
by Eqs. (13) and (21) comes out to be: asymptotic stability of the considered system:
𝑉𝑏𝑎𝑡 𝐿2 𝜃23 1 2 𝛽2 2
𝑢23 = − (53) 𝑉3 = 𝑠 + 𝜓2 (64)
𝑥6 𝑥6 2 3 2
𝑉𝑢𝑐 𝐿3 𝜃45 Taking time derivative of Lyapunov function candidate 𝑉3 yields:
𝑢45 = − (54)
𝑥6 𝑥6
𝑉̇ 3 = 𝑠3 𝑠̇ 3 + 𝛽2 𝜓2 𝜓̇ 2 (65)
Now by using Eqs. (51) and (52), state differential Eqs. (27)–(29) can
be rewritten as: Moreover, Simplifying the equation by substituting the values from Eqs.
𝑑𝑥4 𝑅 (62) and (63a) gives:
= − 2 𝑥4 + 𝜃23 (55) [
𝑑𝑡 𝐿2 𝑉𝑏𝑎𝑡 ( 𝑉 𝑥 )
𝑑𝑥5 𝑅3 𝑉̇ 3 =𝑠3 𝑥4 + 𝑠𝑐 𝑥5 − 6 𝐼𝑜 + 𝛾
= − 𝑥5 + 𝜃45 (56) 𝑥6 𝐶𝑑𝑐 𝑉 𝑉𝑏𝑎𝑡
𝑑𝑡 𝐿3 ] 𝑏𝑎𝑡
− 𝑥̇ 6𝑟𝑒𝑓 + 𝛽2 𝜓2 (66)
( )
𝑑𝑥6 𝐿 𝜃 𝑉 𝑥4 𝐼
= − 2 23 + 𝑏𝑎𝑡 − 𝑜 As the reference value across the DC bus is constant hence its time
𝑑𝑡 𝑥6 𝑥6 𝐶𝑑𝑐 𝐶𝑑𝑐
( ) derivative comes out to be zero. Now by substituting 𝑥̇ 6𝑟𝑒𝑓 = 0 results:
𝐿 𝜃 𝑉 𝑥5
+ − 3 45 + 𝑢𝑐 [ ]
𝑉𝑏𝑎𝑡 ( )
(57)
𝑥6 𝑥6 𝐶𝑑𝑐 𝑉 𝑥
𝑉̇ 3 = 𝑠3 𝑥4 + 𝑠𝑐 𝑥5 − 6 𝐼𝑜 + 𝛾 + 𝛽2 𝜓2 (66a)
𝑥6 𝐶𝑑𝑐 𝑉𝑏𝑎𝑡 𝑉𝑏𝑎𝑡
It can be visualized from Eq. (57) that the nonlinearities involved
in the system are segregated in this equation representing a com- Now to make 𝑉̇ 3 ≤ 0, we choose 𝑥4 as:
plete nonlinear system. Furthermore, to control the SC current, a ( )
𝑉 𝑥 𝑥 𝐶 𝑥 𝐶
proportional–integral (PI) controller is applied as the system is no more 𝑥4 = − 𝑠𝑐 𝑥5 + 6 𝐼𝑜 − 6 𝑑𝑐 𝜙𝑠3 − 6 𝑑𝑐 𝛽2 𝜓2
𝑉𝑏𝑎𝑡 𝑉𝑏𝑎𝑡 𝑉𝑏𝑎𝑡 𝑉𝑏𝑎𝑡
than the combination of linear subsystems. The state equation involved
for DC bus voltage, directly depends upon the battery current i.e. state − 𝜎𝑠𝑔𝑛(𝑠3 ) (67)
𝑥4 . Therefore, to design a robust integral backstepping controller the The terms inside the parenthesis represents nominal control whereas
state Eqs. (55) and (57) are used. Now simplifying the state Eq. (57) the term with signum function denotes switching control. By placing
results: the value of 𝑥4 from Eq. (67) in Eq. (66a) gives:
[
𝑑𝑥6 𝑉 𝑉 𝑥 [ (
= 𝑏𝑎𝑡 𝑥4 + 𝑢𝑐 𝑥5 − 6 𝐼𝑜 𝑉𝑏𝑎𝑡 𝑉 𝑥 𝑉 𝑥 𝑥 𝐼 𝑥 𝐼
𝑑𝑡 𝑥6 𝐶𝑑𝑐 𝑉𝑏𝑎𝑡 𝑉𝑏𝑎𝑡 𝑉̇ 3 = 𝑠3 − 𝑠𝑐 5 + 𝑠𝑐 5 − 6 𝑜 + 6 𝑜
] 𝑥6 𝐶𝑑𝑐 𝑉𝑏𝑎𝑡 𝑉𝑏𝑎𝑡 𝑉𝑏𝑎𝑡 𝑉
𝐿 𝜃 𝐿 𝜃 ) 𝑏𝑎𝑡 ]
− 2 23 𝑥4 − 3 45 𝑥5 (58) 𝑥6 𝐶𝑑𝑐 𝑥6 𝐶𝑑𝑐
𝑉𝑏𝑎𝑡 𝑉𝑏𝑎𝑡 − 𝜙𝑠3 − 𝛽 𝜓 − 𝜎𝑠𝑔𝑛(𝑠3 ) + 𝛾 + 𝛽2 𝜓2 (68)
𝑉𝑏𝑎𝑡 𝑉𝑏𝑎𝑡 2 2
To observe the behavior of the controller under bounded uncertainties
and for the ease of calculation 𝛾 is defined as: [ ]
𝑉𝑏𝑎𝑡 𝑥 𝐶
𝑉̇ 3 = 𝑠3 − 6 𝑑𝑐 𝜙𝑠3 + 𝛾 − 𝜎𝑠𝑔𝑛(𝑠3 ) (68a)
𝐿2 𝜃23 𝐿 𝜃 𝑥6 𝐶𝑑𝑐 𝑉𝑏𝑎𝑡
𝛾=− 𝑥 − 3 45 𝑥5 (59)
𝑉𝑏𝑎𝑡 4 𝑉𝑏𝑎𝑡
[ ]
This equation provides input variations in the sliding surface and is 𝑉𝑏𝑎𝑡 |𝑥 |𝐶
𝑉̇ 3 ≤ |𝑠3 | − 6 𝑑𝑐 𝜙|𝑠3 | + |𝛾| − |𝜎|𝑠𝑔𝑛(𝑠3 ) (69)
quite important when dealing with sliding mode controller. In con- |𝑥6 |𝐶𝑑𝑐 𝑉𝑏𝑎𝑡
clusion, the state dynamics of PHEV, incorporating the predefined If |𝜎| ≥ |𝛾|, the stability from the Lyapunov function can be guaranteed
disturbance is modeled in the compact form as: and performing algebraic simplifications the above equation takes the
𝑑𝑥4 𝑅 form:
= − 2 𝑥4 + 𝜃23
𝑑𝑡 𝐿2
𝑑𝑥5 𝑅3 𝑉̇ 3 = −𝜙𝑠23 (70)
=− 𝑥 + 𝜃45
𝑑𝑡 𝐿3 5 The stability of the Lyapunov function guarantees the perfect volt-
[ ]
𝑑𝑥6 𝑉 𝑉 𝑥 age regulation on the DC bus. Furthermore, the signum function used
= 𝑏𝑎𝑡 𝑥4 + 𝑢𝑐 𝑥5 − 6 𝐼𝑜 + 𝛾 (60)
𝑑𝑡 𝑥6 𝐶𝑑𝑐 𝑉𝑏𝑎𝑡 𝑉𝑏𝑎𝑡 in Eq. (69) produces unnecessary chattering, thus an alternate i.e. sat-
Now to ensure the finite time convergence, the sliding mode surface is uration function is incorporated. The reason for the specified use of sat
defined as: function is its gradually elevated response defined as:
𝑠3
𝑠3 = 𝑥6 − 𝑥6𝑟𝑒𝑓 (61) 𝑠𝑔𝑛(𝑠3 ) = (71)
|𝑠3 | + 𝜇3

6
Z.e. Huma et al. Journal of Energy Storage 42 (2021) 103079

where 𝜇3 defines the isolated operating region of the PHEV. More-

over, the Lyapunov function is stabilized only if the errors associ-
ated with battery and supercapacitor current minimizes to zero. The
corresponding errors are detailed here:

𝑒3 = 𝑥4 − 𝑥4𝑟𝑒𝑓 (72)
𝑒4 = 𝑥5 − 𝑥5𝑟𝑒𝑓 (73)

To minimize the errors 𝑒3 and 𝑒4 to zero to ensure current tracking of

the power sources, following PI controllers are used:

𝑢23 = 𝑘𝑝1 𝑒3 + 𝑘𝑖1 𝑒3 𝑑𝑡 (74)

∫

𝑢45 = 𝑘𝑝2 𝑒4 + 𝑘𝑖2 𝑒4 𝑑𝑡 (75)

∫
( )
𝑉 𝑥 𝑥 𝐶 𝑥 𝐶
𝑥4 = − 𝑠𝑐 𝑥5 + 6 𝐼𝑜 − 6 𝑑𝑐 𝜙𝑠3 − 6 𝑑𝑐 𝛽2 𝜓2
𝑉𝑏𝑎𝑡 𝑉𝑏𝑎𝑡 𝑉𝑏𝑎𝑡 𝑉𝑏𝑎𝑡
𝑠3
−𝜎 (76)
|𝑠3 | + 𝜇3
In conclusion, the reference of the battery current can hence be
generated by using a robust integral backstepping controller and is
notified by Eq. (76). In addition, for simulation workouts, an observer
need not be designed for the state variables 𝑥4 (battery current) and
𝑥5 (SC current), because they can be measured directly using physical
instruments. Moreover, to avoid human error, state observers are to be
designed but with a trade-off in terms of increased computational cost.

4. Supervisory control (rule based strategy) Fig. 4. Rule based energy management strategy of PHEV.

The flow chart for the supervisory control of PHEV has been de-
picted in Fig. 4. This rule-based strategy ensures the power balance for (2) Static Mode (𝑃𝑙𝑜𝑎𝑑 = 0) In this mode, the vehicle is considered
the HESS and a unified charger. A satisfactory response on active and to be at rest with no load requirement. It needs to be charged if
SoC of the battery is less than 10%. This results in the power of
reactive power balancing for the energy storage unit has been ensured,
the main source to be necessarily zero (𝑃𝑏𝑎𝑡 = 0). SC possesses
by considering a limit on SoC of the respective storage elements.
the high power density; hence it will be charged through the
Following targets are achieved by using a rule-based strategy for the excessive energy of the battery and regenerative braking. The
energy management of the system: primary source is plugged in to charge until its SoC reaches
the predefined limit of 90%. After this safe limit, the vehicle is
• Enhancing the life of storage devices (battery and SC) by limiting capable to maneuver according to the load requirements. The
their operations within the defined SoC. parallel combination of SC with the battery in the energy storage
• Incorporating the secondary source during high load demands to unit helps in reducing the stress from the primary source in a
reduce the power accent from the primary source. remarkable pattern. The smooth execution of supervisory control
• Smooth execution of power distribution strategy during regener- helps in the increased life and performance of the system.
ative braking and the transient load.
Furthermore, to observe the SoC of power sources Coulomb counting
method is used and is defined in the following set of equations [44]:
Moreover, depending upon the drive train and power delivering ability
of the primary source, two modes of operation are defined as: 1
𝑆𝑜𝐶𝐵𝐴𝑇 = 𝑆𝑜𝐶𝑏𝑎𝑡 − 𝑖 𝑑𝑡 (77)
3600𝐶𝑁 ∫ 𝑏𝑎𝑡
(1) Dynamic Mode (𝑃𝑙𝑜𝑎𝑑 > 0) 1
𝑆𝑜𝐶𝑆𝐶 = 𝑆𝑜𝐶𝑠𝑐 − 𝑖 𝑑𝑡 (78)
This mode dedicates the vehicle to move according to the speci- 3600𝐶𝑆𝐶 ∫ 𝑠𝑐
fied load profile. The initial condition for the system to start-up where, 𝑖𝑏𝑎𝑡 and 𝑖𝑠𝑐 represents the present current of battery and super-
depends upon the behavior of the load profile. The requisite capacitor respectively. Whereas, 𝑆𝑜𝐶𝐵𝐴𝑇 and 𝑆𝑜𝐶𝑆𝐶 shows the current
power will be supplied mutually by the battery and the SC SoC of the storage elements. Also, the terms 𝑆𝑜𝐶𝑏𝑎𝑡 and 𝑆𝑜𝐶𝑠𝑐 are the
depending upon their power characteristics. The SC being the initial state of charge of power sources. Likely, 𝐶𝑁 and 𝐶𝑆𝐶 are the
auxiliary source will compensate the power transients whereas nominal storage capacities of battery and SC respectively. The Coulomb
the normal load is catered by the battery. During the startup counting method used has been implemented using the given set of
condition, both the battery and the SC will supply the power equations (Eqs. 77 and 78). Furthermore, to estimate the SoC of battery
to balance the load requirements. Moreover, in the case of SC and SC the ideal system in the simulation work with no real time
if SoC<50% then it is charged through regenerative braking. losses has been considered and hence reducing the self-discharge risk.
Conversely, the HESS will continue to supply the demanding In addition, for the real-time implementation of this system this factor
needs to be involved.
power to the motors unless the SoC of the battery falls below
10%. After this certain limit, the battery is unable to fulfill the 5. Simulation results and analysis
load deficit. Additionally, to enhance the operating range of
the energy storage unit, the elements used in HESS need to be In this section, the performance of the proposed nonlinear controller
designed optimally [43]. has been verified using MATLAB/Simulink testbench. The parameters

7
Z.e. Huma et al. Journal of Energy Storage 42 (2021) 103079

Table 1
PHEV specifications.
Vehicle specifications
Car mass (M) 1066 Kg
Front area (A) 1.8 m2
Aerodynamics drag coefficient (𝐶𝑥 ) 0.19
Rolling resistance coefficient (𝐶𝑟 ) 0.0048
DC bus voltage (𝑉𝐷𝐶𝑏𝑢𝑠 ) 400 V
Inverter efficiency (𝜂𝑖𝑛𝑣 ) 75%
Smart charging source
RMS grid nominal voltage 240 V
Output DC voltage 24 V
Rated power 1.5 KW
Frequency 50–60 Hz
Efficiency 85%

Table 2
Battery and supercapacitor parameters.
Battery parameters
Battery type Lithium-Ion (Li-Ion)
Terminal voltage 190–230 V
Rated current capacity 13.9 Ah
Capacity per cell 1.4 Ah
Battery array parallel: 25, series: 64
Supercapacitor parameters
Nominal voltage 205 V
Total capacitance 2700 F
Initial SoC 1
Current threshold −80–80 A
Supercapacitor array parallel: 20, series: 20

Fig. 5. (a) Battery current 𝑖𝐵 (b) Battery-inner voltage 𝑉𝐵 (CC–CV process).

Table 3
Simulation parameters.
Circuit parameters
Inductances 𝐿1 , 𝐿2 , 𝐿3 495 μH, 1 mH, 1 mH
Capacitances 𝐶1 , 𝐶𝑑𝑐 70 nF, 95 mF
Resistances 𝑅1 𝑅2 , 𝑅3 , 0.5 mΩ, 10 mΩ, 10 mΩ
Switching frequency 100 KHz
Controller parameters
𝑘𝑝1 , 𝑘𝑝2 , 𝑘𝑖1 , 𝑘𝑖2 9000, 900, 1000, 200
𝜇1 , 𝜇2 , 𝜇3 1, 1, 400
𝛽1 , 𝛽2 0.8, 10
𝑘1 , 𝑘2 5, 5
𝜎, 𝜙 10, 100

used for the modeling and simulation of the system have been listed
in Tables 1–3. The target of the entire setup is to ensure the stable
Fig. 6. Intermediate DC bus of charger.
operation and efficient performance of the PHEV. The gains used for
the tuning of the proposed nonlinear controller have been obtained
by using a genetic gain estimation-based algorithm. The results ob-
tained from the simulation work consist of vehicle load profile, voltage
regulation at DC bus, current tracking, and SoC of the contributing
sources and are depicted sequentially in Figs. 5–15. The entire section
is split into two subsections. The first subsection represents the results
for the plug side of the vehicle and the second subsection represents the
dynamic behavior of the HESS according to the load profile. Depending
upon the load profile, the second subsection is split into two cases. First
with varying load and the second with EUDC (European Extra-Urban
Drive Cycle) load profile. Moreover, to ensure the effectiveness of the
proposed robust integral backstepping controller, it has been compared
against the control Lyapunov function (CLF) and backstepping (BS) Fig. 7. Case:1 DC load current profile (varying).

controller. The simulation is executed for 𝑡 = 15 s for the varying load

and 𝑡 = 400 s in case of EUDC, with the regulated output voltage of
400𝑉 .

8
Z.e. Huma et al. Journal of Energy Storage 42 (2021) 103079

Fig. 8. Case:1 DC bus voltage during dynamic mode.

Fig. 10. (a) Case:1 SoC of Battery (b) Case:1 SoC of SC.

Fig. 11. Speed profile of vehicle under standard EUDC.

Fig. 9. (a) Case:1 Battery Current (𝑥4 ) (b) Case:1 SC Current (𝑥5 ).

The regulated charging voltage at the intermediate DC bus has been

5.1. Results for nonlinear control of integrated battery charger set at 48𝑉 . In addition, to observe the robustness of the proposed
controller, the results for the charging voltages have been plotted in
The selection of design characteristics for the charger relates di- parallel with CLF and BS controller that can be visualized in Fig. 6.
rectly to the profile of the battery used in PHEV. A battery with a It is pertinent to mention that the proposed nonlinear controller out-
classes the other controllers with no overshoots/undershoots resulting
large storage capacity demands the high-power balancing ability of
in balanced power for the charging unit. Moreover, the behavior of the
the charger. The robust integral backstepping (RIBS) based nonlinear
CLF does not satisfy the requirements as the surge can be observed at
controller has been proposed to stabilize the charging current and the
the initial point, with overshoots/undershoots around the equilibrium
voltages. Fig. 5a illustrates the battery current during the complete
point. Similarly, the backstepping controller leads to the undesired high
stage of constant current (CC) and constant voltage (CV). During the
voltage peak during startup along with oscillations and steady-state
CC stage, the battery current stabilizes its value at 10𝐴 until the battery error throughout the charge fitting line. The supervisory control helps
inner voltage reaches its reference value of 48𝑉 . At this particular in reducing the stress from the charging unit along with enhanced
instant, the CV stage begins and the battery current starts approaching efficiency and battery life.
its minimum value. Fig. 5b represents the battery inner voltage for the
CC–CV cycle. In CC stage the battery inner voltage increases gradually 5.2. Results for nonlinear control of HESS
until it reaches its reference value of 48𝑉 . Then it remains constant for
the complete CV stage until the battery is charged to its safe limit. The 5.2.1. Case-1 (varying load)
whole CC–CV charging process of the Lithium-ion battery is illustrated In this case, the controller is verified for dynamical load conditions.
by Fig. 5a and b. The load current is varied between 30 − 45𝐴 and 45 − 25𝐴 as shown in

9
Z.e. Huma et al. Journal of Energy Storage 42 (2021) 103079

Fig. 12. Case:2 DC load current profile (EUDC).

Fig. 15. (a) Case:2 SoC of Battery (b) Case:2 SoC of SC (EUDC).

Fig. 13. Case:2 DC bus voltage(𝑥6 ) EUDC.

proposed controller has been compared with standard controllers like

CLF and BS. It is clear from the figure that the CLF controller shows a
regular high voltage pulse with some steady-state error during instant
variations in load profile at 𝑡 = 5 s and 𝑡 = 10 s. Similar is the case with
the BS controller which reflects high voltage transients and fluctuations
around the mean value during every rise and fall in load profile. For the
varying load profile, the behavior of the proposed controller is inspiring
in terms of voltage regulation as no transients or steady-state error has
been observed.
Now the Fig. 9a and b have been drawn to observe the current
tracking performance of the proposed controller and behavior of the
energy storage elements. Considering the time interval from 𝑡 = 0 − 5𝑠,
when the vehicle began to move with the considered load current of
30𝐴, both the battery and SC began to discharge to overcome the high
load requirements. During the time 𝑡 = 5−10𝑠 as the load current further
increases from 30𝐴 to 45𝐴 both the sources continue to provide the
deficit current. Finally, during the interval 𝑡 = 10−15𝑠 the load demand
falls to 25𝐴 resulting in reduced stress on the energy storage unit.
Moreover, during this span, stress from the primary source is reduced
and hence the SC begins to charge from the regenerative braking. The
energy management algorithm helps in reducing stress and enhancing
the life of the power-up sources. The effectiveness of the proposed
algorithm has been guaranteed by visualizing the results for SoC of both
the sources presented by Fig. 10a and b.

5.2.2. Case-2 (EUDC)

The proposed controller has also been tested for the vehicle moving
under standard EUDC. It is considered to be a standardized profile
to estimate the operating efficiency and behavior of the vehicle. The
Fig. 14. (a) Case:2 Battery Current (𝑥4 ) (b) Case:2 SC Current (𝑥5 ).
speed profile and load current 𝐼𝑜 associated with the EUDC have been
presented by Figs. 11 and 12. The driving load profile is generated by
using the following basic equation:
Fig. 7. The proposed robust integral backstepping controller tracks the [ ]
voltage at DC bus in just 0.02𝑠 and is depicted by Fig. 8. The dynamic 1 𝑑𝑣𝑡
𝐼𝑜 = 0.5𝜌𝑎 𝑣2𝑡 𝐴𝐶𝑥 + 𝑀𝑔 𝐶𝑟 𝑀 𝑣𝑡 (79)
response of the controller is effective for the varying load profile. The (𝜂𝑖𝑛𝑣 )(𝑉𝐷𝐶𝑏𝑢𝑠 ) 𝑑𝑡

10
Z.e. Huma et al. Journal of Energy Storage 42 (2021) 103079

Fig. 17. DC bus voltage (state 𝑥6 ).

Fig. 16. C-HIL test bench for HESS.

where 𝜂𝑖𝑛𝑣 represents the inverter efficiency and 𝑉𝐷𝐶𝑏𝑢𝑠 shows the DC
bus voltage, 𝑣𝑡 represents the vehicle speed, 𝑀 and 𝐴 depicts the mass
and front area of the car respectively. Moreover, 𝐶𝑥 and 𝐶𝑟 are the
coefficients of the aerodynamics drag and rolling resistance respectively
with 𝜌𝑎 as the opposing air density. The operational efficiency of the
inverter is assumed to be 75%. The parameters utilized in Eq. (78) to
approximate the load current 𝐼𝑜 of the vehicle under EUDC have been
detailed in Table 1 [42].
Furthermore, to observe the regulation of the bus voltage at the
designated value and to test the effectiveness of the proposed ro-
bust integral backstepping controller, it has been compared with some
state-of-the-art controllers. It is worth mentioning that the proposed
controller tracks the DC bus voltage at 400𝑉 in just 0.01𝑠. Whereas,
Fig. 18. (a) Battery current(𝑥4 ) (b) SC current (𝑥5 ).
CLF and BS controllers achieve the desired reference in 30𝑠 and 10𝑠
respectively as depicted in Fig. 13. Furthermore, CLF and BS controllers
are quite rich in overshoots/undershoots. Now Fig. 14a and b have been
drawn to observe the current tracking performance of the proposed has been established to ensure the robustness of the proposed controller
controller and behavior of the power-up sources. Initially, both the for PHEVs. Fig. 16 shows the real-time workstation for C-HIL. The
battery and SC discharges at the rapid rate to fulfill the high load simulated controller has been evaluated experimentally using C2000
requirements. At 𝑡 = 70 s, as the load demand stabilizes and no more Delfino MCU F28379xD LaunchPad to generate the discrete control
transients occur, the primary source is capable enough to cope up with signals with the plant model to be simulated on MATLAB/Simulink.
the load profile. The phenomenon keeps on repeating according to the Moreover, the results obtained from C-HIL has been plotted against the
variation in the load current. Some undershoots can be observed at simulation results to validate the real-time applicability of the proposed
𝑡 = 200 s and 𝑡 = 360 s due to the instant variations in the load controller for the assumed topology.
profile but are well in the safe limit of operation mentioned in Table 1. Fig. 17 shows the regulated voltage at the DC bus with the dynamic
The employed supervisory control results in reduced stress from the load profile of the PHEV. The vehicle is assumed to maneuver with
main and auxiliary sources as shown by the SoC graphs in Fig. 15a varying speed, resulting in the dynamic load current. It is pertinent to
and b. As our desired control targets have been achieved, assuring the mention that the result obtained for the regulated output voltage is in
effectiveness of the proposed controller. well accordance with the simulation results. Although minute fluctua-
tions have been observed during the load variations but are well in the
6. Hardware in the loop verification safe limit of operation. The plotted DC bus profile reflects few spikes
when there is an increase or decrease in the load current at 𝑡 = 45 s and
The simulation results obtained by MATLAB/Simulink reflect the 𝑡 = 55 s respectively. In a similar pattern, the charging and discharging
efficient response of the system and the proposed controller, but several current profiles of HESS have been plotted against the software results
impulsive factors affect the behavior of the PHEV in real-time applica- as depicted in Fig. 18a and b. The robustness of the proposed controller
tion. Thus, the real-time controller hardware in the loop (C-HIL) setup has been guaranteed by the absolute voltage regulation of the DC bus

11
Z.e. Huma et al. Journal of Energy Storage 42 (2021) 103079

and the current tracking of the power sources. The real-time C-HIL [8] S. Kumar, A. Usman, A review of converter topologies for battery charging ap-
results indicate the similar behavior of the proposed RIBS controller plications in plug-in hybrid electric vehicles, in: 2018 IEEE Industry Applications
Society Annual Meeting, IAS, IEEE, 2018, pp. 1–9.
when simulated on MATLAB/Simulink, justifying the effective response
[9] A. Rachid, H. El Fadil, F. Giri, A. Lassioui, Nonlinear output feedback control
of the controller. of V2G single-phase on-board BEV charger, Asian J. Control 22 (5) (2020)
1848–1859.
7. Conclusion [10] M.S. Khan, I. Ahmad, F.Z.U. Abideen, Output voltage regulation of FC-UC based
hybrid electric vehicle using integral backstepping control, IEEE Access 7 (2019)
65693–65702.
In this research paper a new arrangement for the PHEVs constituting
[11] N. Denis, M.R. Dubois, R. Dubé, A. Desrochers, Blended power management
the integrated charger, battery and supercapacitor, has been proposed strategy using pattern recognition for a plug-in hybrid electric vehicle, Int. J.
to enhance the life and deliverance ability of the battery as a primary Intell. Transp. Syst. Res. 14 (2) (2016) 101–114.
source. Robust integral backstepping-based nonlinear controller has [12] H. El Fadil, F. Giri, J.M. Guerrero, A. Tahri, Modeling and nonlinear control of a
been proposed for the smart execution and hybrid energy management fuel cell/supercapacitor hybrid energy storage system for electric vehicles, IEEE
Trans. Veh. Technol. 63 (7) (2014) 3011–3018.
of the PHEV. The controller has been designed for the regulation
[13] M.A. Majeed, M.G. Khan, F. Asghar, Nonlinear control of hybrid energy storage
of output voltage of an integrated charger and DC bus with current system for hybrid electric vehicles, Int. Trans. Electr. Energy Syst., e12268.
tracking of the storage devices to their respective references. The [14] M.F. Munir, I. Ahmad, S.A. Siffat, M.A. Qureshi, H. Armghan, N. Ali, Non-linear
energy management algorithm comprising of power balance, has been control for electric power stage of fuel cell vehicles, ISA Trans. (2020).
used for the efficient response of the system. The detailed analysis [15] Z. Song, J. Hou, H. Hofmann, J. Li, M. Ouyang, Sliding-mode and Lyapunov
using Lyapunov stability criteria has been done and the simulation function-based control for battery/supercapacitor hybrid energy storage system
used in electric vehicles, Energy 122 (2017) 601–612.
results have been presented using MATLAB/Simulink which guarantee
[16] R.K. Sharma, S. Mishra, Dynamic power management and control of a PV PEM
the asymptotic stability of the system. Furthermore, to ensure the fuel-cell-based standalone ac/dc microgrid using hybrid energy storage, IEEE
effectiveness of the proposed controller, it has been compared with con- Trans. Ind. Appl. 54 (1) (2017) 526–538.
trol Lyapunov function and backstepping based nonlinear controllers [17] M. El-Sharkh, A. Rahman, M. Alam, P. Byrne, A. Sakla, T. Thomas, A dynamic
proposed in the literature. The documented results reflect the improved model for a stand-alone PEM fuel cell power plant for residential applications,
J. Power Sources 138 (1–2) (2004) 199–204.
and satisfactory response of the proposed controller. The achieved
[18] N. Sulaiman, M. Hannan, A. Mohamed, E. Majlan, W.W. Daud, A review on
results show better performance of the proposed controller with no energy management system for fuel cell hybrid electric vehicle: Issues and
overshoots/undershoots and finite time tracking during instant load challenges, Renew. Sustain. Energy Rev. 52 (2015) 802–814.
variations. To ensure the real-time compatibility of the controller, it [19] A.U. Rahman, I. Ahmad, A.S. Malik, Variable structure-based control of fuel cell-
has been implemented experimentally using controller hardware in the supercapacitor-battery based hybrid electric vehicle, J. Energy Storage 29 (2020)
101365.
loop setup. The experimental behavior depicts that the controller is
[20] H. Armghan, I. Ahmad, N. Ali, M.F. Munir, S. Khan, A. Armghan, Nonlinear
capable of swamping the transients and oscillations around the equi- controller analysis of fuel cell-battery-ultracapacitor-based hybrid energy storage
librium point. The future contribution may emphasize on the type and systems in electric vehicles, Arab. J. Sci. Eng. 43 (6) (2018) 3123–3133.
performance of the power devices using different control techniques. [21] D. Xu, Q. Liu, W. Yan, W. Yang, Adaptive terminal sliding mode control for
Different nonlinear controllers incorporating adaptation laws can also hybrid energy storage systems of fuel cell, battery and supercapacitor, IEEE
Access 7 (2019) 29295–29303.
be designed to make the system robust against disturbances and noise.
[22] A. Khaligh, Z. Li, Battery, ultracapacitor, fuel cell, and hybrid energy storage
Model-based predictive controllers can also be used to estimate the load
systems for electric, hybrid electric, fuel cell and plug-in hybrid electric vehicles:
profile of the motor. Moreover, in HESS the fuel cell or PV system State of the art, IEEE Trans. Veh. Technol. 59 (6) (2010) 2806–2814.
can be used as a continuous power source to maintain the constant [23] H. Armghan, M. Yang, M. Wang, N. Ali, A. Armghan, Nonlinear integral
energy level as required by the vehicle. Moreover, in case of a decline backstepping based control of a DC microgrid with renewable generation and
in the behavior of the system, the compensation mechanism might be energy storage systems, Int. J. Electr. Power Energy Syst. 117 (2020) 105613.
[24] F. Naseri, E. Farjah, T. Ghanbari, An efficient regenerative braking system based
presented.
on battery/supercapacitor for electric, hybrid and plug-in hybrid electric vehicles
with BLDC motor, IEEE Trans. Veh. Technol. 66 (5) (2016) 3724–3738.
Declaration of competing interest [25] M. Vajedi, N.L. Azad, Ecological adaptive cruise controller for plug-in hybrid
electric vehicles using nonlinear model predictive control, IEEE Trans. Intell.
Transp. Syst. 17 (1) (2015) 113–122.
The authors declare that they have no known competing finan-
[26] B. Yang, T. Zhu, X. Zhang, J. Wang, H. Shu, S. Li, T. He, L. Yang, T. Yu,
cial interests or personal relationships that could have appeared to
Design and implementation of battery/SMES hybrid energy storage systems used
influence the work reported in this paper. in electric vehicles: A nonlinear robust fractional-order control approach, Energy
191 (2020) 116510.
References [27] F. Akar, Y. Tavlasoglu, E. Ugur, B. Vural, I. Aksoy, A bidirectional nonisolated
multi-input DC–DC converter for hybrid energy storage systems in electric
vehicles, IEEE Trans. Veh. Technol. 65 (10) (2015) 7944–7955.
[1] D. Dominković, I. Bačeković, A.S. Pedersen, G. Krajačić, The future of transporta-
tion in sustainable energy systems: Opportunities and barriers in a clean energy [28] J. Li, Q. Yang, F. Robinson, F. Liang, M. Zhang, W. Yuan, Design and test of a
transition, Renew. Sustain. Energy Rev. 82 (2018) 1823–1838. new droop control algorithm for a SMES/battery hybrid energy storage system,
[2] M. Dreidy, H. Mokhlis, S. Mekhilef, Inertia response and frequency control Energy 118 (2017) 1110–1122.
techniques for renewable energy sources: A review, Renew. Sustain. Energy Rev. [29] X. Lu, J.M. Guerrero, K. Sun, J.C. Vasquez, An improved droop control method
69 (2017) 144–155. for dc microgrids based on low bandwidth communication with dc bus voltage
[3] M. Gejguš, C. Aschbacher, J. Sablik, Comparison of the total costs of renewable restoration and enhanced current sharing accuracy, IEEE Trans. Power Electron.
and conventional energy sources, Res. Pap. Fac. Mater. Sci. Technol. Slovak Univ. 29 (4) (2013) 1800–1812.
Technol. 24 (37) (2016) 99–104. [30] J.-B. Cao, B.-G. Cao, Fuzzy-logic-based sliding-mode controller design for
[4] S. Zhang, R. Xiong, J. Cao, Battery durability and longevity based power position-sensorless electric vehicle, IEEE Trans. Power Electron. 24 (10) (2009)
management for plug-in hybrid electric vehicle with hybrid energy storage 2368–2378.
system, Appl. Energy 179 (2016) 316–328. [31] Z. Bai, Z. Yan, X. Wu, J. Xu, B. Cao, H∞ control for battery/supercapacitor
[5] B.M. Al-Alawi, T.H. Bradley, Review of hybrid, plug-in hybrid and electric vehicle hybrid energy storage system used in electric vehicles, Int. J. Automot. Technol.
market modeling studies, Renew. Sustain. Energy Rev. 21 (2013) 190–203. 20 (6) (2019) 1287–1296.
[6] C.-S.N. Shiau, C. Samaras, R. Hauffe, J.J. Michalek, Impact of battery weight [32] L. Wang, Y. Wang, C. Liu, D. Yang, C. Zonghai, A power distribution strategy
and charging patterns on the economic and environmental benefits of plug-in for hybrid energy storage system using adaptive model predictive control, IEEE
hybrid vehicles, Energy Policy 37 (7) (2009) 2653–2663. Trans. Power Electron. (2019).
[7] S. Lacroix, E. Labouré, M. Hilairet, An integrated fast battery charger for electric [33] Q. Li, W. Chen, Y. Li, S. Liu, J. Huang, Energy management strategy for fuel
vehicle, in: 2010 IEEE Vehicle Power and Propulsion Conference, IEEE, 2010, cell/battery/ultracapacitor hybrid vehicle based on fuzzy logic, Int. J. Electr.
pp. 1–6. Power Energy Syst. 43 (1) (2012) 514–525.

12
Z.e. Huma et al. Journal of Energy Storage 42 (2021) 103079

[34] B. Yang, J. Wang, X. Zhang, J. Wang, H. Shu, S. Li, T. He, C. Lan, T. Yu, [39] H.-C. Chang, C.-M. Liaw, Development of a compact switched-reluctance motor
Applications of battery/supercapacitor hybrid energy storage systems for electric drive for EV propulsion with voltage-boosting and PFC charging capabilities,
vehicles using perturbation observer based robust control, J. Power Sources 448 IEEE Trans. Veh. Technol. 58 (7) (2009) 3198–3215.
(2020) 227444. [40] M. Yilmaz, P.T. Krein, Review of battery charger topologies, charging power
[35] S. Ahmed, H.M. Muhammad Adil, I. Ahmad, M.K. Azeem, S. Abbas Khan, et al., levels and infrastructure for plug-in electric and hybrid vehicles, IEEE Trans.
Supertwisting sliding mode algorithm based nonlinear MPPT control for a solar Power Electron. 28 (5) (2012) 2151–2169.
PV system with artificial neural networks based reference generation, Energies [41] A.M. Haidar, K.M. Muttaqi, Behavioral characterization of electric vehicle charg-
13 (14) (2020) 3695. ing loads in a distribution power grid through modeling of battery chargers,
[36] B. Wang, J. Xu, R.-J. Wai, B. Cao, Adaptive sliding-mode with hysteresis control IEEE Trans. Ind. Appl. 52 (1) (2015) 483–492.
strategy for simple multimode hybrid energy storage system in electric vehicles, [42] M.S. Khan, I. Ahmad, H. Armaghan, N. Ali, Backstepping sliding mode control
IEEE Trans. Ind. Electron. 64 (2) (2016) 1404–1414. of FC-UC based hybrid electric vehicle, IEEE Access 6 (2018) 77202–77211.
[37] S. East, M. Cannon, Energy management in plug-in hybrid electric vehicles: [43] J. Shen, S. Dusmez, A. Khaligh, Optimization of sizing and battery cycle life
Convex optimization algorithms for model predictive control, IEEE Trans. Control in battery/ultracapacitor hybrid energy storage systems for electric vehicle
Syst. Technol. (2019). applications, IEEE Trans. Ind. Inf. 10 (4) (2014) 2112–2121.
[38] X. Zhou, G. Wang, S. Lukic, S. Bhattacharya, A. Huang, Multi-function bi- [44] H. Mahmood, D. Michaelson, J. Jiang, A power management strategy for
directional battery charger for plug-in hybrid electric vehicle application, PV/battery hybrid systems in islanded microgrids, IEEE J. Emerg. Sel. Top. Power
in: 2009 IEEE Energy Conversion Congress and Exposition, IEEE, 2009, pp. Electron. 2 (4) (2014) 870–882.
3930–3936.

13