Sustainable Energy Technologies and Assessments 50 (2022) 101880

Contents lists available at ScienceDirect

Sustainable Energy Technologies and Assessments

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

Linear quadratic regulator controllers for regulation of the dc-bus voltage in

a hybrid energy system: Modeling, design and experimental validation
Majid A. Abdullah a, *, Ali Q. Al-Shetwi b, M. Mansor c, M.A. Hannan c, Chee Wei Tan d, A.H.
M. Yatim d
a
University of Hafr Al Batin, King Abdulaziz Rd, Al Baladiyah, Hafar Al Batin 3992, Saudi Arabia
b
Department, Fahad Bin Sultan University, Tabuk 47721, Saudi Arabia
c
Institute of Power Engineering, Universiti Tenaga Nasional, Kajang 43000, Malaysia
d
Department of Electrical Power Engineering, School of Electrical Engineering, Faculty of Engineering Universiti Teknologi Malaysia, 81300 Skudai, Johor, Malaysia

A R T I C L E I N F O A B S T R A C T

Keywords: In this paper, linear quadratic regulator (LQR) controllers for effective operation of a hybrid energy system
Linear quadratic regulator (LQR) consisting of ultracapacitor energy storage and wind energy system have been designed and implemented. The
Wind energy conversion system (WECS) control objective is to regulate the dc-bus voltage to a target level while extracting the maximum power from the
Ultracapacitor
available wind. The dc-bus voltage regulation is achieved by controlling the charging and discharging of the
Energy storage system
Bidirectional converter
ultracapacitor through a bidirectional converter, and tracking the maximum power points (MPPs) is achieved by
Boost converter controlling a boost converter interfacing the wind turbine with the dc-bus. In addition, a boost converter-based
wind turbine emulator to behavior similar to a real wind generator has been developed for testing the proposed
controllers. The performance of the proposed energy system incorporating the LQR controllers has been tested
under several scenarios (both in simulations and experiments), and the results presented. The simulation tests
were conducted in the environment of MATLAB/Simulink, and the experimental tests implemented based on
low-cost Digital Signal Processor (DSP) TMS320F2812 eZdsp board. The simulation and experimental results
demonstrate their consistency and the capability of the proposed LQR controllers to (1) track the reference
voltages and currents, and (2) swiftly recover the nominal operating condition of the system at all conditions
including any variation in wind speed or load demand.

Introduction turbines necessarily varies with the change of wind speed [4]. In this
sense, a dc-dc boost converter with a maximum power point tracking
The world is witnessing increasing energy demand as a result of a (MPPT) capability is needed to optimize the wind turbine intermittent
high energy consumption rate and the rapid growth of population. To energy generation [5]. Numerous algorithms for tracking maximum
fulfill this growing demand, most countries depend on fossil-fueled power points (MPPs) of WECSs have been developed in recent years. A
power plants for electric power generation. However, because of the study by Kumar and colleagues [6] provides a systematic review for
many negative effects arising from burning fossil fuels, the alternative many of these algorithms. Because of the simplicity and fast tracking,
and promising solution is to replace these traditional sources with clean the One-Power-Point (OPP) MPPT algorithm is selected for tracking the
and renewable energy sources. Among all different renewable energy MPPs in the WECS in this paper. Because of space limitation, the details
sources, wind energy, including small wind energy systems, has a of the OPP MPPT algorithm are omitted here; for more information on
promising potential nowadays because of its many advantages, such as the algorithm’s concept and implementation, readers may refer to
sustainability, environmental friendliness, and low production costs, Abdullah’s study [7].
and less maintenance [1–3]. However, the output power of wind The MPPT is achieved by continuously adjusting the rotor speed

Abbreviations: WECS, Wind Energy Conversion System; LQR, Linear Quadratic Regulator; MPPs, Maximum Power Points; MPPT, Maximum Power Point Tracking;
OPP, One-Power-Point; PI, Proportional Integral; P&O, Perturbation and Observation.
* Corresponding author.
E-mail addresses: [email protected] (M.A. Abdullah), [email protected] (A.Q. Al-Shetwi), [email protected] (M. Mansor), hannan@uniten.
edu.my (M.A. Hannan), [email protected] (C.W. Tan), [email protected] (A.H.M. Yatim).

https://doi.org/10.1016/j.seta.2021.101880
Received 20 March 2021; Received in revised form 23 September 2021; Accepted 11 December 2021
Available online 21 December 2021
2213-1388/© 2021 Elsevier Ltd. All rights reserved.
M.A. Abdullah et al. Sustainable Energy Technologies and Assessments 50 (2022) 101880

Fig. 1. Schematic diagram of the ultracapacitor interfaced to the WECS.

according to the instantaneous variations in wind speed. However, control law for optimal performance of linear multivariable systems
because the output voltage of the WECS is proportional to the rotor [24]. The control performance of LQR was compared to that of the PI
speed, changing the rotational speed is accompanied by a change in controller in a study by Abdullah et al. [25] for the case of regulating
output voltage, leading to a variable dc-bus voltage. Consequently, to output voltage of a boost converter. Based on the comparison, the LQR
maintain a constant dc-bus voltage despite the variations in wind speed, controller was found to achieve better performance in terms of settling
WECS is often interfaced with an energy storage system composed of a time as well as over- and under-shoots. The LQR controller performance
storage device and a bidirectional dc-dc converter. The energy storage in achieving MPPT of a stand-alone photovoltaic system was also
device improves the system resiliency by absorbing excess energy during compared to the traditional Perturbation and Observation (P&O)
periods of surplus energy generation or injecting it into the load during method by Anbarasi and Kanthalakshmi [26]. It was found that the LQR
periods of insufficient energy [8,9]. The bidirectional converter controls method performs better than the traditional P&O method based on real-
both charging and discharging of the storage device, by transferring the time implementation using the dSPACE controller. In addition, Deihimi
energy between the WECS and the storage device in both directions and Mahmoodieh [27] and Fard and Aldeen [28] applied the LQR
[10]. Compared to other storage devices, the ultracapacitor is preferred method to successfully regulate the output voltage and power of a
because of its low initial cost, better electrical performance, and long hybrid PV and battery energy system. The LQR control method was
calendar life cycle. In addition, the ultracapacitor has the benefits of chosen because of its many advantages, such as robustness, ability to
high power density and fast and efficient charge/discharge capability supply the energy of the control signals, and ability to systematically
[9]. Moreover, ultracapacitors have no internal chemical reactions, do calculate the integral state-feedback gain matrix [27]. According to the
not require heating or cooling, have no moving parts, and are robust and findings of Fard and Aldeen study [28], the designed controller could
more efficient [11]. satisfy the objectives while demonstrating acceptable undershoot and
In general, integrating renewable energy sources and storage devices good performance. Similarly, in a work by Batiyah et al. [29], a stand-
requires power converters controlled by robust and stable control sys­ alone and grid-connected PV system were controlled by LQR-based
tems to ensure reliable and efficient operation of the hybrid energy voltage controllers for tracking the MPPs during grid-connected mode
systems. An effective control method will not only have a good and regulating the dc-bus voltage during stand-alone operation. Ac­
smoothing effect, but also can reduce the size of the storage device and cording to the simulation results, the control objectives were achieved
save investment costs for users [12]. Proportional integral (PI) con­ with a good transient performance under different test cases. In another
trollers are the classical and most widely-used controllers designed for study by M Mendis et al. [30], LQR-based controllers were designed and
controlling these converters, as described in [13,14]. The key feature of simulated for controlling the dc-dc converters interfacing a hybrid
PI controllers is that they are less complex and hence easier to imple­ microgrid consisting of a solar PV system, a battery energy storage
ment [15]; however, they have several drawbacks. According to litera­ system, and a diesel generator with a rectifier. The performance of the
ture, PI controllers have the following major drawbacks: sensitivity to proposed LQR controllers was then evaluated against PI controllers.
systems with highly non-linear components [16], poor dynamic Comparative results indicate that the fluctuations in dc-bus voltage are
response [17], weak disturbance rejection [18], and lack of robustness lower using the LQR controllers relative to those using the PI controllers,
against the variation of the operating points [19,20]. Alternatively, the which again proves the superiority of the LQR controllers. Aryani et al.
linear quadratic regulator (LQR) optimal control method offers good [31] used LQR-based current control for an interlink bidirectional ac/dc
control qualities, including attractive gain and phase margins [21] as converter in a hybrid microgrid with ac and dc subgrids. In comparison
well as converters’ robustness over a wide operational range [22,23]. In to the PI controller, the LQR demonstrated higher performance and
addition, LQR control provides a systematic approach to design the robust control under the various operating circumstances examined

2
M.A. Abdullah et al. Sustainable Energy Technologies and Assessments 50 (2022) 101880

[31]. Similarly, based on a technical overview of different control

methods for distributed generation units in an islanded microgrid, the
authors Hossain et al. [32] concluded that the LQR is positively char­
acterized by rapid dynamic response, accurate tracking ability, and a
relatively simple designing procedure when compared to the other
control methods reviewed.
Because of the aforementioned advantages, in this paper, the appli­
cation of the LQR control method for controlling the power electronic
converters integrating WECS and an ultracapacitor storage system to a
common dc-bus is investigated. Two LQR feedback controllers are
designed in this work to ensure the robustness of the system, improve
reference tracking, and optimize the closed loop performance of the
bidirectional and boost dc-dc converters used in the integration. These
Fig. 2. Boost converter circuit topology.
controllers are designed based on obtaining the optimal state-feedback
gains that minimize the weighted linear combination of the converter
accomplished by controlling the boost converter switch’s duty-cycle.
states and duty cycle described in the LQR cost function. By using this
The dc-bus voltage is regulated by controlling the duty-cycle of the
control technique, the converters’ stability is guaranteed over a wide
bidirectional converter switches. The bidirectional dc-dc converter
range of operating conditions and in both buck and boost modes of the
transmits energy in both directions and is functionally similar to two dc-
bidirectional converter. Initial results of this work have been presented
dc converters: buck and boost converters. However, instead of imple­
in [33–35]; an extension of the previous work with experimental vali­
menting individual controllers for buck charging and boost discharging
dation is included in this paper. The extension is the application of the
current modes, it is preferable to realize a unified controller that drives
LQR controllers designed in [33] for input current regulation and in [34]
the converter switches by a set of complementary signals because a
for output voltage regulation into the hybrid energy system simulated in
unified controller simplifies the design and implementation while
[35]. In addition, the major contribution of this paper is the experi­
ensuring smooth mode transition capability [34]. In this paper, the boost
mental validation of the DSP-based LQR controllers incorporated into a
mode of operation is selected to design the bidirectional converter
developed hybrid energy system that can track MPPs while ensuring the
controller.
voltage across the dc-bus is maintained at the target reference. The
performance of the proposed hybrid energy system with LQR controllers
is verified by simulation in MATLAB/Simulink and validated by con­ Boost converter model
ducting experimental tests over a wide range of operating conditions.
Based on the obtained simulation and experimental results, the designed Fig. 2 shows the boost converter circuit. The first step in designing
LQR controllers proved their capability to track the reference current the LQR control for this circuit is to obtain its state-space model. The
and voltage with good dynamic performance. This paper is organized as model is derived using a well-known converter modeling technique
follows: Section System description” presents an overview on the stud­ called the state-space averaging method. Based on [33], the boost con­
ied hybrid wind/ultracapacitor energy system. Next, a LQR control is verter state-space model is represented as:
applied to regulate the dc-bus voltage and extracting maximum power { ′ ′
̂
from the WECS in Section “LQR controller design”. Hardware design and ẋ = A ̂
⌢
x+B ̂
′
u
′ (1)
implementation are then provided in Section “Hardware design and ẏ = Ĉx+E ̂ u
implemetation”. Detailed validation results are presented in Section
“Results and discussions”, and the conclusions are presented in Section where:
“Conclusion”. ⎡ ⎤
(D − 1)R2 Ro (D − 1)Ro
⎢ (R2 + Ro )L 0
⎢ (R2 + Ro )L ⎥
⎥
System description ⎢ ⎥
⎢ − 1 ⎥
A=⎢⎢ 0 0 ⎥
⎥ (2)
R1 Cin
The block diagram of the proposed hybrid wind-ultracapacitor en­ ⎢
⎢
⎥
⎥
ergy system implemented in this work is depicted in Fig. 1. The WECS is ⎣ (1 − D)Ro
0
− 1 ⎦
comprised of a variable speed permanent magnet synchronous generator (R2 + Ro )Co (R2 + Ro )Co
(PMSG)-based wind turbine connected to the dc-bus through a power ⎡ ⎤
electronic converter. This power converter is composed of a diode bridge Ro (R2 IL + Vo ) 1 (D − 1)R2 Ro
⎢ (R + R )L (R2 + Ro )L ⎥
rectifier connected to a dc-dc boost converter with MPPT capability. The ⎢ 2 o L ⎥
⎢ ⎥
wind turbine converts the kinetic energy in the wind into mechanical ⎢ 1 ⎥
3)
′
B =⎢ 0 0 ⎥
power. The PMSG then converts this mechanical power into usable ⎢
⎢ R1 Cin ⎥
⎥
electrical power. The boost converter maximizes the extracted electrical ⎢
⎣ − Ro Ro
⎥
⎦
power by forcing the energy system to track the maximum power points. (R2 + Ro )Co
IL 0
(R2 + Ro )Co
The ultracapacitor and the bidirectional converter constitute the energy
storage system. The ultracapacitor smoothes the fluctuating output [
(1 − D)R2 Ro Ro
]
power from the WECS by injecting or absorbing higher frequency power C= 0 (4)
R2 + Ro R2 + Ro
fluctuations. The bidirectional converter regulates the dc-bus voltage by
controlling the charging and discharging of the ultracapacitor. The [
R2 Ro
]
(5)
′
bidirectional converter has two terminals, namely, the low voltage side E = 0 0
R2 + Ro
and the high voltage side. The ultracapacitor is connected to the low
voltage side, and the high voltage side is connected to the dc-bus. x = [ ̂i L
̂ ̂v Cin ̂v Co ] (6)

LQR controller design

(7)
′
̂ = [̂
u d v in ̂i o ]
̂
Tracking the MPPs in the studied system represented in Fig. 1 is

3
M.A. Abdullah et al. Sustainable Energy Technologies and Assessments 50 (2022) 101880

Fig. 4. The structure of the proposed LQR-based controller for regulating the
output voltage of the boost converter.
Fig. 3. The structure of the proposed LQR-based controller for regulating the
input current of the boost converter.

Regulation of input current

The input current controller is designed to ensure that the converter

input current follows accurately a desired reference trajectory while
exhibiting the robustness against both input voltage fluctuations and
load changes. For a good tracking, the open-loop system is augmented
by adding an integral action to the current tracking error, as shown in
Fig. 3. Thus, the new state-space variable,ξ = ̂
x 4 , is defined as:

x 4 = ̂i ref − ̂i L = ̂i ref − Ĉ
ξ̇ = ̂ x (8)

C = [1 0 0] (9)
With inclusion this new state variable, the state vector then becomes
xa = [ ̂
̂ x ̂ ξ ]. As a result, the extended converter model is defined as:
[ ] [ ] [ ′ ][ ′ ]
A3×3 03×1 x
̂ B 3×3 03×3 u
̂
̂
ẋ a = ̂ + (10) Fig. 5. Experimental setup for wind-ultracapacitor hybrid energy system.
− C1×3 01×1 ξ 01×3 I1×3 ̂i ref
⏟̅̅̅̅̅̅̅̅̅̅̅̅⏞⏞̅̅̅̅̅̅̅̅̅̅̅̅⏟Aa
the voltage error signal is created within the controller and is used as a
where matrices A, B and ̂ u are as prescribed in (2), (3), and (7), feedback term, as shown in Fig. 4. Because the bidirectional converter
′ ′

respectively. The control variable ̂

d is defined based on the feedback has no input capacitor, no state variable is related to the input capacitor.
gain K and the extended state vector ̂
x a as: Thus, the new state variable becomes:

̂
d = − K̂
x a = [ − k1 − k2 − k3 − k4 ]̂
xa (11) x 3 = ̂v ref − ̂v o = ̂v ref − Ĉ
ξ̇ = ̂ x (13)

The optimal LQR controller is obtained by minimizing the following C = [0 1] (14)

quadratic cost function:
The converter augmented model is written as:
∫ +∞
2 [ ] [ ] [ ′ ][ ′ ]
(12)
T
J= x a + ρ̂
x a Q̂
(̂ d )dt A2×2 0 x
̂ B 3×3 03×1 u
̂
0
̂
ẋ a = ̂ + (15)
− C1×2 0 ξ 01×3 1 ̂v ref
[ ] ⏟̅̅̅̅̅̅̅̅̅̅⏞⏞̅̅̅̅̅̅̅̅̅̅⏟Aa
I3×3 03×1
where ρ > 0 and Q = ⩾0.
01×3 q ⎡
− (1 − D)
⎤
Selection of the values of weighting coefficients q and ρ has a crucial ⎢
0
L ⎥
effect on the control performance. It is important to choose these values A=⎢
⎣1 − D
⎥
⎦ (16)
− 1
carefully to minimize the cost function and also to yield the desired Co Ro C o
response. The appropriate values are often determined based on expe­
rience and primarily through the process of trial and error. However, The minimized LQR cost function in this case is similar to that given
according to [33], a large value for q should be chosen to eliminates the by (12), except that ̂ x and Q are now defined as:
steady-state error and guarantee the best possible integrator perfor­ [ ]
I 02×1
mance. In contrast, the value of ρ is recommended to be small to achieve x = [ ̂i L ̂v Co ]Q = 2×2
̂
01×2 q
fast disturbance rejection. Once appropriate values of ρ and q are
selected, the LQR optimal controller design reduces to computing the
Hardware design and implemetation
vector of state-feedback gains K that minimizes the quadratic cost
equation (12).
A photograph of the hardware test setup and the schematic diagram
of the complete system implementation are shown in Figs. 5 and 6,
respectively. In the experimental setup, the converter’s controllable
Regulation of output voltage
power switches (S, S1 , S2 and S3 ) are 40-A/60-V IGBTs G20N60B3D,
and the inductors (L, L1 and L2 ) are Hammond commercially manufac­
The output voltage controller is designed to ensure that the converter
tured 1 mH 10A type. The used capacitors are 470 μF of the electrolytic
output voltage follows accurately a desired reference trajectory while
type, and the ultracapacitor is BOOSTCAP BMOD0165E48B (Maxwell
achieving the robustness against both input voltage fluctuations and
Technologies). The load is emulated using a programmable dc electronic
load changes. Thus, a new state variable that computes the integral of

4
M.A. Abdullah et al. Sustainable Energy Technologies and Assessments 50 (2022) 101880

Fig. 6. Schematic diagram of the hardware configuration.

load. The IGBTs gate signals are controlled by a Texas Instruments (TI) configuration is shown as parts of Figs. 6 and 7. As shown in Fig. 6, the
TMS320F2812 DSP board utilizing IGBT driver boards. boost converter circuit’s input terminal is connected to the power sup­
ply, while its output terminal is connected to the second boost converter
circuit representing the MPPT circuit. The mathematical models of the
Real-time wind generator emulator turbine and the control system are implemented in the DSP, as shown in
Fig. 7. By controlling the converter output voltage (Vdc ), the same
The Performance evaluation of the designed DSP-based LQR con­ electrical behavior of the real wind turbine is generated. The control
trollers requires a real-time emulator that can reproduce precisely the action is achieved using the duty ratio of the switch (S). The proposed
same non-linear output voltage that a real wind turbine generator pro­ emulator, which is inexpensive, simple-to-operate, and reconfigurable,
duces in a natural environment. Such a real-time emulator is tradition­ can be used for prototype testing and design purposes for wind energy-
ally constructed using a real DC or AC motor coupled with a real based generation systems.
generator. However, such emulators are characterized by their high
cost, complexity, and time-consuming implementation. Besides, we only
need the wind turbine’s electrical characteristics for testing the pro­ LQR controllers implementation
posed controllers, not the mechanical ones. For this reason, an
economical, simple, and flexible emulator based on a mathematical Fig. 7 shows the implemented DSP-based LQR controllers block di­
model of the WECS, a dc-dc boost converter, and a power supply is agram. As illustrated in the figure, three LQR-based controllers were
developed. The schematic diagram of the proposed emulator’s hardware developed, each of which controls a single dc-dc converter, as briefly

5
M.A. Abdullah et al. Sustainable Energy Technologies and Assessments 50 (2022) 101880

Boost conv.1 controller

Vw

Wind Turbine
Idc &
PMSG models
IL IL_ADC S

Vdc-ref
IL
LQR d
Vdc Controller 1

Idc Idc_ADC

Boost conv.2 controller S1

MPPT Algorithm
Signals Conditioning
A/D Converter

PWM
Vdc Vdc_ADC

I dc - ref

d1 S2
LQR
Controller 2
Vo Vo_ADC Vo

IL2 d2
LQR
Controller 3
IL2 IL2_ADC d3 S3

Vo - ref
Bidirectional conv. controller

Fig. 7. The DSP-based controllers block diagram.

Fig. 8. Process of implementing the model.

clarified in the subsequent subsections. The steps of implementation of Composer Studio to be loaded and executed on the DSP board.
the DSP-based control system are summarized in Fig. 8. A Simulink
model of the controllers is firstly developed in MATLAB/Simulink using LQR controller 1
the Embedded Target Blocks for TI C2000 DSP. The Simulink model is The objective of this controller is to guarantee that the output voltage
then converted by Real-Time Workshop to C programming code. Sub­ of the boost converter 1 tracks precisely the reference voltage generated
sequently, the generated code is compiled and linked using Code by the wind turbine model (Vdc\_ref ). This voltage reference emulates the

6
M.A. Abdullah et al. Sustainable Energy Technologies and Assessments 50 (2022) 101880

output voltage of a real wind generator and reflects the variation in wind
speed. The controller is fed with the sensed output voltage and inductor
current of the boost converter 1 to generate the corresponding duty ratio
that is sent to the PWM block to generate the switching pulses for switch
S.

LQR controller 2
The objective of this controller is to guarantee that the input current
of the boost converter 2 (Idc ) tracks precisely the reference current
(Idc\_ref ) obtained by the MPPT algorithm. The controller is fed with the
sensed input current (Idc ), input voltage (Vdc ), and output voltage (Vo ) of
the boost converter 2 to generate the corresponding duty ratio (d1 ) that
is sent to the PWM block to generate the switching pulses for switch S1 .

LQR controller 3
The aim of this controller is to ensure that the dc-bus voltage tracks
the desired reference value (Vo\_ref ) precisely. Tracking the desired
reference is accomplished by controlling the charging and discharging of
the ultracapacitor. This controller is fed with the output voltage and
inductor current of the bidirectional converter (Vo\_ref and IL1 , respec­
tively). The outputs from the controller are the duty ratios (d3 and d4 ).
These complementary signals are sent to the PWM block to generate the
switching pulses for switches S3 and S4 .

Results and discussions

The complete hybrid energy system incorporating the LQR control­

lers was modeled and simulated in MATLAB/Simulink. To significantly
decrease the simulation time, the average-circuit models of the rectifier-
PMSG and the boost and bidirectional converters were used for simu­
lation for this paper. Details on the converters average-circuit models, Fig. 9. The voltage response to the variation in wind speed: (a) simulation and
the wind turbine characteristics, and the ultracapacitor model were (b) experimental results.
discussed in [35]. After evaluating the dynamic responses of the
designed controllers through simulations, the simulation results were
then validated by conducting several laboratory tests. The simulation
and laboratory experiments were subjected to the same test conditions
to obtain a fair comparison of the results.
Note that the weighting matrix Q and weighting factor ρ were tuned
by means of simulation-based examinations in such a manner that
achieves efficient reference tracking, with acceptable over- and under-
shoots as well as acceptable rise and settling times. The corresponding
optimal state-feedback gain vector for the boost converter 2 controller
was calculated with ρ = 10 and with Q given in (17). The resulted gain
vector is expressed in (18).
⎡ ⎤
1 × 10− 3 0 0 0
⎢ 0 1 × 10− 3 0 0 ⎥
Q=⎢ ⎣
⎥ (17)
0 0 1 × 10− 3 0 ⎦
0 0 0 50
[ ]
K = 1.66 × 10− 2
− 1.033 × 10− 3
1.35 × 10− 3
− 2.23 (18)

For both boost converter 1 and the bidirectional converter, a good

transient response was achieved with ρ = 100 and with Q expressed in
(19). The gain vector is shown in (20).
⎡ ⎤
0.01 0 0
Q=⎣ 0 0.01 0 ⎦ (19)
0 0 4
[ ]
K 1 = 0.36 × 10− 3
2.7 × 10− 3
− 0.2 (20)

To evaluate the effectiveness of the designed controllers, the pro­

posed ultracapacitor-wind hybrid energy system incorporating the LQR
controllers was tested under three different operating conditions. The
first case is carried out to test the designed controllers for regulation of
the dc-bus voltage while minimizing the impact of the wind speed Fig. 10. The current response to the variation in wind speed: (a) simulation
and (b) experimental results.

7
M.A. Abdullah et al. Sustainable Energy Technologies and Assessments 50 (2022) 101880

Fig. 11. The voltage response to the variation in dc-bus reference voltage: (a)
simulation and (b) experimental results. Fig. 12. The current response to the variation in dc-bus reference voltage: (a)
simulation and (b) experimental results.
changes. In the second case, the test is carried out to confirm the ability
of the designed controllers to track a prescribed reference voltage while value set by the MPPT algorithm. This increment in IL was accompanied
maintaining the system operation at MPPs. In the third case, the purpose by a drop in the current delivered by the ultracapacitor (IL1 ), ensuring
is to test the tracking performance under load variations. the load current remains constant. In addition, note that the ultra­
capacitor current was always positive, making the bidirectional con­
Case 1: test of wind speed change verter function as a boost converter throughout the whole time of the
simulation.
As aforementioned, the aim of this test is to evaluate the performance
of the LQR controllers in maintaining the dc-bus voltage at the pre­ Case 2: test of DC-Bus reference voltage change
scribed reference under the circumstances of the variations in wind
speed. Thus, for this test, the wind speed was changed at t = 10 s from This test is performed to evaluate the effectiveness of the proposed
5.5 m/s to 6 m/s, and the load resistance and reference dc-bus voltage LQR controllers in forcing the system to track the prescribed changes in
were set fixed at 43 Ω and 60 V, respectively. The system dynamic the dc-bus reference voltage. Thus, unlike the previous case, in this case,
response under step change in wind speed is shown in Figs. 9 and 10 for the wind speed was set constant at 5.5 m/s while the dc-bus reference
voltage and current waveforms, respectively. voltage was varied at t = 10 s from 50 V to 60 V. The load resistance was
According to the voltage waveforms in Fig. 9, the proposed LQR set as the previous case at 43 Ω. Referring to the voltage waveforms in
voltage controller is clearly able to regulate the dc-bus voltage (Vo ) at Fig. 11, a response similar to case 1 is observed throughout the build-up
the target reference value, regardless of wind speed variations, thereby process. As noted, the output voltage of wind generator (Vdc ) could be
demonstrating the controller’s robustness and its sufficient dynamic and successfully emulated using a boost converter controlled by LQR-based
steady-state performance. From the same figure, note the role of ultra­ controller (controller 1). The applied change to dc-bus reference voltage
capacitor in the designed system throughout the build-up process; (10 V) was tracked accurately because of the use of LQR voltage
although the output voltage of the wind generator (Vdc ) required controller 3; the response was found to be fast (approximately 0.2 s) and
approximately 3 s to stabilize, the voltage of the dc-bus stabilized at the without any over- or under-shoots. Furthermore, this figure also proves
reference voltage (60 V) within only 0.3 s. the robustness of LQR controller 1, which was capable to quickly recover
This quick response was achieved by the presence of the ultra­ Vdc to its nominal operating point.
capacitor, which ensures the load is supplied with a constant current Correspondingly, in Fig. 12, the current response of the hybrid en­
(approximately 1.4 A) all the time. According to the current waveform in ergy system is shown. In contrast to the preceding case, which illustrated
Fig. 10, the whole load current (Io ) was supplied by the ultracapacitor at a single mode of operation, this case demonstrates the importance of the
the beginning of the build-up process; after the wind generator current ultracapacitor in allowing the bidirectional converter to work in both
(IL ) began to rise, both currents contributed to supply the load demand buck and boost modes. As noted in the figure, starting at t = 3 s, the wind
the rest of the time. In the same figure, it can be observed that increasing generator current (IL ) was able to cover the full load demand (1 A) with
the wind speed at t = 10 s resulted in an increment in the current of the some extra current used to charge the ultracapacitor through operating
wind generator. This behavior is caused by the presence of LQR the converter in buck mode. Nevertheless, because of the increase of the
controller 2, which forces the WECS system to follow the new reference reference value of the dc-bus voltage (at t = 10 s), the load demand (Io )

8
M.A. Abdullah et al. Sustainable Energy Technologies and Assessments 50 (2022) 101880

smooth transition between the ultracapacitor charging and discharging

modes was ensured through the robust control of the bidirectional
converter using LQR controller 3. As shown in the figure, a good
response was achieved, exhibiting nearly no over- or under-shoots.

Case 3: test of load change

In this case, the performance of the proposed LQR controllers was

tested under load change condition. A step change from 100 Ω to 60 Ω at
t = 10 s was applied on the commercial programmable electronic load
interfaced with the dc-bus. The wind speed and dc-bus reference voltage
were kept constant at 5.5 m/s and 60 V, respectively. It is clear from the
voltage waveforms shown in Fig. 13 that, despite 40% change in load
resistance, the dc-bus voltage recovered to its steady-state value very
quickly with almost negligible transient response. This observation is
further proof of the robust performance of the LQR controllers.
Similar to the previous case, the current waveforms drawn in Fig. 14
show that, in this case, the bidirectional converter operated in buck
mode to charge the ultracapacitor during the first interval (t < 10 s),
where the power delivered by the WECS exceeds that required by the
load. However, once the load resistance changed, the load current
increased to 1 A, which is equal to the wind generator current; hence,
the ultracapacitor charging current dropped to zero. Considering all of
the cases examined in this work, it is evident that the proposed LQR
controllers have been tested under various ultracapacitor operating
conditions, including charging, discharging, and zero current. In all the
tested cases, the proposed controllers have proven their capability and
robustness.
Fig. 13. The voltage response to the variation in load: (a) simulation and (b)
experimental results. Conclusion

The successful design and DSP-based implementation of LQR con­

trollers for an integrated ultracapacitor and wind energy converter
system was presented. Three LQR controllers were utilized in the pro­
posed hybrid system. The first controller was used to ensure extraction
of the maximum power from the wind turbine by means of operating the
WECS at the MPPs at all times. The second controller was used for
maintaining the dc-bus voltage fixed at its desired value, by controlling
the ultracapacitor’s charge and discharge process. The third controller
was used to control a boost dc-dc converter to emulate the output
voltage of wind generator. The proposed LQR controllers were tested
under different simulation and experimental conditions, proving their
robustness and capability in obtaining good dynamic and steady-state
response. These tests were carried out by introducing different step
variations in wind speed, dc-bus reference voltage, and load resistance.
In all the tests, the results showed that the dc-bus voltage was success­
fully controlled to accurately track the reference voltage, obtaining a
short transient period (<0.3 s) without oscillation. In addition, the
proposed LQR controllers demonstrated good tracking ability of the
current of the WECS to the reference current set by the MPPT algorithm.
Moreover, the proposed controllers enabled emulation of a real wind
generator through a simple boost converter circuit, in which the con­
verter output voltage variation is similar to that of the real wind
generator under the same conditions. Overall, it is evident that LQR
controllers can be used to effectively control hybrid energy systems.

Declaration of Competing Interest

The authors declare that they have no known competing financial

interests or personal relationships that could have appeared to influence
Fig. 14. The current response to the variation in load: (a) simulation and (b) the work reported in this paper.
experimental result.

was raised to approximately 1.4 A. To compensate the shortage in load

current, the bidirectional converter changed its operation mode to boost
mode, allowing discharging of the ultracapacitor to the load. The

9
M.A. Abdullah et al. Sustainable Energy Technologies and Assessments 50 (2022) 101880

References [18] Manandhar U, Tummuru NR, Kollimalla SK, Ukil A, Beng GH, Chaudhari K.
Validation of faster joint control strategy for battery-and supercapacitor-based
energy storage system. IEEE Trans Ind Electron 2018;65(4):3286–95.
[1] Xu Q, Xiao J, Wang P, Pan X, Wen C. A decentralized control strategy for
[19] Zhao H, Wu Q, Rasmussen CN, Blanke M. L1 adaptive speed control of a small wind
autonomous transient power sharing and state-of-charge recovery in hybrid energy
energy conversion system for maximum power point tracking. IEEE Trans Energy
storage systems. IEEE Trans Sustainable Energy 2017;8(4):1443–52.
Convers 2014;29(3):576–84.
[2] Ajirlo KS, Tari PH, Gharali K, Zandi M. Development of a wind turbine simulator to
[20] Kassem AM, Yousef AM. Voltage and frequency control of an autonomous hybrid
design and test micro HAWTs. Sustainable Energy Technol Assess 2021;43:100900.
generation system based on linear model predictive control. Sustainable Energy
[3] Mishra J, Pattnaik M, Samanta S. Drift-free perturb and observe mppt algorithm
Technol Assess 2013;4:52–61.
with improved performance for SEIG-based stand-alone wind energy generation
[21] Patarroyo-Montenegro JF, Andrade F, Guerrero JM, Vasquez JC. A linear quadratic
system. IEEE Trans Power Electron 2019;35(6):5842–9.
regulator with optimal reference tracking for three-phase inverter-based islanded
[4] Tiwari SK, Singh B, Goel PK. Control of wind-diesel hybrid system with BESS for
microgrids. IEEE Trans Power Electron 2020;36(6):7112–22.
optimal operation. IEEE Trans Ind Appl 2019;55(2):1863–72.
[22] Quan X, Wu Z, Dou X, Hu M, Huang AQ. Load current decoupling based LQ control
[5] Fathy A, Alharbi AG, Alshammari S, Hasanien HM. Archimedes optimization
for three-phase inverter. IEEE Trans Power Electron 2018;33(6):5476–91.
algorithm based maximum power point tracker for wind energy generation system.
[23] Arab N, Kedjar B, Javadi A, Al-Haddad K. A Multifunctional single-phase grid-
Ain Shams Eng J 2021.
integrated residential solar PV systems based on LQR control. IEEE Trans Ind Appl
[6] Kumar MBH, Saravanan B, Sanjeevikumar P, Blaabjerg F. Review on control
2019;55(2):2099–109.
techniques and methodologies for maximum power extraction from wind energy
[24] Cheema MAM, Fletcher JE, Xiao D, Rahman MF. A linear quadratic regulator-based
systems. IET Renew Power Gener 2018;12(14):1609–22.
optimal direct thrust force control of linear permanent-magnet synchronous motor.
[7] Abdullah MA, Al-Hadhrami T, Tan CW, Yatim AH. Towards green energy for smart
IEEE Trans Ind Electron 2016;63(5):2722–33.
cities: particle swarm optimization based MPPT approach. IEEE Access 2018;6:
[25] Abdullah M, Tan C, Yatim A. A simulation comparison of PI and Linear Quadratic
58427–38.
Regulator controllers in DC-DC converter. In: 2015 IEEE Conference on Energy
[8] Bhosale R, Agarwal V. Fuzzy logic control of the ultracapacitor interface for
Conversion (CENCON). IEEE; 2015. p. 37–41.
enhanced transient response and voltage stability of a dc microgrid. IEEE Trans Ind
[26] Mp A, Kanthalakshmi S. Linear quadratic optimal control of solar photovoltaic
Appl 2019;55(1):712–20.
system: an experimental validation. J Renewable Sustainable Energy 2016;8.
[9] Rahimzadeh A, Christiaanse TV, Evins R. Optimal storage systems for residential
[27] Deihimi A, Mahmoodieh MES. Analysis and control of battery-integrated dc/dc
energy systems in British Columbia. Sustainable Energy Technol Assess 2021;45:
converters for renewable energy applications. IET Power Electron 2017;10(14):
101108.
1819–31.
[10] Pires VF, Foito D, Cordeiro A. A DC–DC converter with quadratic gain and
[28] Fard M, Aldeen M. Linear Quadratic Regulator design for a hybrid photovoltaic-
bidirectional capability for batteries/supercapacitors. IEEE Trans Ind Appl 2018;54
battery system. In: 2016 Australian Control Conference (AuCC). IEEE; 2016.
(1):274–85.
p. 347–52.
[11] Faisal M, Hannan MA, Ker PJ, Hussain A, Mansor MB, Blaabjerg F. Review of
[29] Batiyah S, Zohrabi N, Abdelwahed S, Qunais T, Mousa M. Optimal control design of
energy storage system technologies in microgrid applications: issues and
a voltage controller for stand-alone and grid-connected PV converter. In: 2018
challenges. IEEE Access 2018;6:35143–64.
IEEE Texas Power and Energy Conference (TPEC). IEEE; 2018. p. 1–6.
[12] Zhang D, et al. Control strategy and optimal configuration of energy storage system
[30] Mendis N, Mahmud M, Roy T, Haque ME, Muttaqi KM. Power management and
for smoothing short-term fluctuation of PV power. Sustainable Energy Technol
control strategies for efficient operation of a solar power dominated hybrid DC
Assess 2021;45:101166.
microgrid for remote power applications. In: 2016 IEEE Industry Applications
[13] Xiao J, Wang P, Setyawan L. Multilevel energy management system for
Society Annual Meeting. IEEE; 2016. p. 1–8.
hybridization of energy storages in DC microgrids. IEEE Trans Smart Grid 2016;7
[31] Aryani DR, Kim J-S, Song H. Interlink converter with linear quadratic regulator
(2):847–56.
based current control for hybrid AC/DC microgrid. Energies 2017;10(11):1799.
[14] Korada N, Mishra MK. Grid adaptive power management strategy for an integrated
[32] Hossain MJAJA, Pota HR, Issa W, Hossain MJAJA. Overview of AC microgrid
microgrid with hybrid energy storage. IEEE Trans Ind Electron 2017;64(4):
controls with inverter-interfaced generations. Energies 2017;10(9):1300.
2884–92.
[33] Al-Mothafar M, Radaideh S, Abdullah M. LQR–based control of parallel–connected
[15] Lim YS, Wong J, Liew MSS, Khaw LYA. Proportional integrator (PI) and fuzzy-
boost dc–dc converters: a comparison with classical current–mode control. Int J
controlled energy storage for zero-power flow between grid and local network with
Comput Appl Technol 2012;45(1):15–27.
photovoltaic system. Sustainable Energy Technol Assess 2020;37:100629.
[34] Abdullah M, Yatim A, Tan C, Samosir A. Control of a bidirectional converter to
[16] Housseini B, Okou F, Beguenane R. Robust nonlinear controller design for on-grid/
interface ultracapacitor with renewable energy sources. In: 2013 IEEE
off-grid wind energy battery-storage system. IEEE Trans Smart Grid 2017.
International Conference on Industrial Technology (ICIT). IEEE; 2013. p. 673–8.
[17] Zhang X, Wang B, Manandhar U, Gooi HB, Foo G. A model predictive current
[35] Abdullah M, Tan C, Yatim A. A simulation study of hybrid wind-ultracapacitor
controlled bidirectional three-level dc/dc converter for hybrid energy storage
energy conversion system. In: 2014 IEEE Conference on Energy Conversion
system in dc microgrids. IEEE Trans Power Electron 2018;34(5):4025–30.
(CENCON). IEEE; 2014. p. 265–70.

10