RNN-INTEGRATED MODEL PREDICTIVE CONTROL FOR FUEL CELL AND SOLAR-POWERED HYBRID ELECTRIC VEHICLES
This paper presents an innovative Hybrid Electric Vehicle (HEV) configuration utilizing a fuel cell as the primary energy source and an onboard Photovoltaic (PV) array as a supplementary source. The system features an advanced Model Predictive Control (MPC) enhanced by a Recurrent Neural Network (RN...
Gespeichert in:
| Datum: | 2024 |
|---|---|
| Hauptverfasser: | , |
| Format: | Artikel |
| Sprache: | Englisch |
| Veröffentlicht: |
Institute of Renewable Energy National Academy of Sciences of Ukraine
2024
|
| Schlagworte: | |
| Online Zugang: | https://ve.org.ua/index.php/journal/article/view/483 |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| Назва журналу: | Vidnovluvana energetika |
| Завантажити файл: |  |
Institution
Vidnovluvana energetika| _version_ | 1871103820030803968 |
|---|---|
| author | Divya, G. Venkata Padmavathi , S. |
| author_facet | Divya, G. Venkata Padmavathi , S. |
| author_institution_txt_mv | [
{
"author": " G. Divya",
"institution": "GITAM School of Technology, GITAM Deemed to be University, Hyderabad, Telangana; EEE, CVR College of Engineering, Hyderabad, Telangana "
},
{
"author": "S. Venkata Padmavathi ",
"institution": "EEE, GITAM School of Technology, GITAM Deemed to be University, Hyderabad, Telangana"
}
] |
| author_sort | Divya, G. |
| baseUrl_str | https://ve.org.ua/index.php/journal/oai |
| collection | OJS |
| datestamp_date | 2026-07-18T06:32:20Z |
| description | This paper presents an innovative Hybrid Electric Vehicle (HEV) configuration utilizing a fuel cell as the primary energy source and an onboard Photovoltaic (PV) array as a supplementary source. The system features an advanced Model Predictive Control (MPC) enhanced by a Recurrent Neural Network (RNN) to manage the induction motor efficiently. Key components include a PV array, a fuel cell, and an electrolyzer. The PV array supplements the fuel cell during optimal sunlight conditions, while excess energy during idle periods is converted to hydrogen via the electrolyzer and stored in a hydrogen tank for future use. A quadratic bidirectional buck-boost converter (QBBC) regulates voltage, ensuring compatibility between energy sources and the motor. The system’s performance is evaluated under various sunlight and speed conditions, with the RNN-based MPC compared to an Artificial Neural Network-based MPC (ANN-MPC) and a traditional Proportional-Integral (PI) controller. An incremental conductance algorithm is implemented for Maximum Power Point Tracking (MPPT) to optimize PV power extraction. The RNN model predicts motor speed, enhancing control precision. Simulations in MATLAB/SIMULINK reveal that the RNN-based MPC outperforms ANN-MPC and PI controllers, demonstrating improved efficiency and speed control. This work contributes to advancing intelligent and energy-efficient HEV technologies. |
| doi_str_mv | 10.36296/1819-8058.2024.4(79).17-28 |
| first_indexed | 2025-07-17T11:39:40Z |
| format | Article |
| fulltext |
17
Відновлювана енергетика. №4/2024 | Комплексні проблеми енергетичних систем на основі НВДЕ
UDK 621 https://doi.org/10.36296/1819-8058.2024.4(79)17-28
RNN-INTEGRATED MODEL PREDICTIVE CONTROL FOR FUEL CELL AND SOLAR-POWERED HYBRID
ELECTRIC VEHICLES
Received Sept. 04, 2024; accepted Nov. 27, 2024
Available online Dec. 11, 2024
Divya G.1, Venkata Padmavathi S.2
Author for correspondence: Divya G.,
e-mail: divyaphd0@gmail.com
Abstract. This paper presents an innovative Hybrid Electric Vehi-
cle (HEV) configuration utilizing a fuel cell as the primary energy
source and an onboard Photovoltaic (PV) array as a supplemen-
tary source. The system features an advanced Model Predictive
Control (MPC) enhanced by a Recurrent Neural Network (RNN)
to manage the induction motor efficiently. Key components in-
clude a PV array, a fuel cell, and an electrolyzer. The PV array
supplements the fuel cell during optimal sunlight conditions, while excess energy during idle periods is con-
verted to hydrogen via the electrolyzer and stored in a hydrogen tank for future use. A quadratic bidirectional
buck-boost converter (QBBC) regulates voltage, ensuring compatibility between energy sources and the mo-
tor. The system’s performance is evaluated under various sunlight and speed conditions, with the RNN-based
MPC compared to an Artificial Neural Network-based MPC (ANN-MPC) and a traditional Proportional-Inte-
gral (PI) controller. An incremental conductance algorithm is implemented for Maximum Power Point Track-
ing (MPPT) to optimize PV power extraction. The RNN model predicts motor speed, enhancing control preci-
sion. Simulations in MATLAB/SIMULINK reveal that the RNN-based MPC outperforms ANN-MPC and PI
controllers, demonstrating improved efficiency and speed control. This work contributes to advancing intelli-
gent and energy-efficient HEV technologies.
Keywords: Hybrid Electric Vehicle (HEV), Recurrent Neural Network (RNN), Fuel Cell, Induction Motor Control,
Electrolyzer, Renewable Energy Integration, Speed Control, MATLAB/SIMULINK Simulation.
І. INTRODUCTION
The global shift towards environmentally sustainable trans-
portation has driven the development of advanced hybrid
and electric vehicle technologies [1]. Hybrid Electric Vehi-
cles (HEVs) have emerged as a promising solution, combin-
ing multiple energy sources to maximize efficiency and re-
duce emissions. The integration of hydrogen fuel cells with
photovoltaic (PV) systems for Hybrid Electric Vehicles
(HEVs) has been extensively studied in recent years [2]. In
[3], authors investigated such integration to improve en-
ergy efficiency, revealing significant performance enhance-
ments. However, they highlighted a major drawback in the
form of increased vehicle weight and space limitations due
to the large hydrogen storage system, impacting the vehi-
cle’s compactness and practicality. Similarly, in [4] authors
examined bidirectional converters to facilitate dynamic
voltage control within hybrid systems. Although effective in
steady states, their approach faced limitations in real-time
response, struggling under rapidly fluctuating load condi-
tions common in real-world driving scenarios. In [5], re-
searchers proposed a Maximum Power Point Tracking
(MPPT) algorithm optimized for PV arrays in electric vehi-
cles. Their technique exhibited good efficiency under stable
irradiance but suffered from performance degradation
under partial shading conditions, reducing energy harvest-
ing capabilities. In [6], authors explored the application of
traditional Proportional-Integral (PI) controllers for HEV
motor drives. While easy to implement, PI controllers
lacked robustness and failed to handle variations in power
input efficiently, especially under fluctuating renewable en-
ergy sources.
Artificial Neural Network-based Model Predictive Control
(ANN-MPC) is proposed in [7] to improve the performance
of electric vehicles. Despite demonstrating enhanced effi-
ciency and control precision, the computational complexity
of ANN-MPC posed a challenge, making real-time applica-
tion impractical. Similarly, in [8], authors focused on fuel
cell technology for hybrid vehicle systems, emphasizing the
benefits of extended range and clean energy. However,
their system's fuel cell stack exhibited low durability under
high load cycles, leading to frequent maintenance and re-
duced lifespan. In [9], authors worked on refining the incre-
mental conductance MPPT algorithm for hybrid systems.
Though effective, their method required frequent tuning to
maintain performance, complicating practical implementa-
tions. In [10], authors researched indirect vector control
techniques for induction motors, achieving significant im-
provements in torque response. Yet, the system struggled
1 Ph. D Scholar, Assistant Professor
https://orcid.org/0000-0003-3320-0060
2 Dr. of Science, Assistant Professor
https://orcid.org/0000-0001-8792-2152
1 GITAM School of Technology, GITAM
Deemed to be University, Hyderabad,
Telangana; EEE, CVR College of
Engineering, Hyderabad, Telangana
2 EEE, GITAM School of Technology, GITAM
Deemed to be University, Hyderabad,
Telangana
18
Відновлювана енергетика. №4/2024 | Комплексні проблеми енергетичних систем на основі НВДЕ
with maintaining performance at lower speeds, impacting
vehicle drivability and overall efficiency. In [11], authors de-
veloped heuristic-based energy management strategies for
balancing power among multiple sources in HEVs. How-
ever, these methods proved suboptimal in handling sudden
and unexpected energy demands. In [12], authors applied
deep learning models to predict motor speeds in electric
vehicles. Their models achieved high accuracy but were
heavily reliant on large volumes of training data and signif-
icant computational resources, limiting their scalability. Bi-
directional DC-DC converters to improve power transfer in
hybrid systems is proposed in [13]. Although achieved ef-
fective power conversion, the system was prone to electro-
magnetic interference, affecting the reliability and stability
of power electronics. In [14], authors explored using ultra-
capacitors for energy storage in HEVs, providing high peak
power. However, they struggled with limited energy den-
sity, making ultracapacitors less suitable for long-distance
travel.
The use of Recurrent Neural Networks (RNNs) for vehicle
control systems for achieving good predictive accuracy is
analysed in [15]. Despite their benefits, RNN models were
found to be sensitive to noise in sensor data, reducing their
reliability under real-world conditions. In [16], authors
tested MPPT techniques for PV optimization in vehicles,
showing improved power extraction. However, their sys-
tem failed to adapt efficiently to rapid changes in vehicle
orientation, such as when driving on uneven terrain. The
optimization of fuel cell efficiency using advanced thermal
management systems is studied in [17]. Their approach sig-
nificantly reduced energy losses, but the added complexity
and cost of the cooling system presented practical imple-
mentation challenges. In [18], authors designed adaptive
control strategies for torque management in HEVs, which
enhanced vehicle performance. Yet, these strategies were
highly sensitive to parameter variations, requiring exten-
sive calibration for each vehicle model. Authors in [19] in-
vestigated using combined solar and wind energy systems
for hybrid vehicles. While this improved energy availability,
the variability and unpredictability of wind energy limited
the system's reliability.
High-efficiency converters in hybrid systems for achieving
low power losses is studied in [20]. Nevertheless, their de-
sign exhibited poor thermal performance, necessitating ad-
vanced cooling methods that increased the overall system
complexity. In [21], researchers presented a hybrid energy
management approach combining rule-based and optimi-
zation-based strategies. Their model performed well under
specific conditions but struggled with computational effi-
ciency when scaling up to more complex scenarios. Predic-
tive torque control for induction motors is proposed in [22],
which enhanced control accuracy but demanded a high
sampling rate, increasing computational overhead. In [23],
authors developed an intelligent energy distribution sys-
tem for HEVs using machine learning algorithms. Although
their system effectively balanced energy usage, it required
continuous data updates and suffered from model drift
over time, limiting long-term reliability.
This paper introduces a novel HEV configuration that har-
nesses the potential of renewable energy and cutting-edge
control systems. The proposed system integrates a hydro-
gen-based fuel cell as the primary energy source, supple-
mented by an onboard Photovoltaic (PV) array. The PV ar-
ray plays a dual role: it provides additional power to
support the fuel cell under optimal sunlight conditions and,
during vehicle idle periods, channels excess energy to an
electrolyzer that generates hydrogen for future use. This
setup not only enhances energy efficiency but also lever-
ages renewable energy for sustainable vehicle operation. A
critical component of this HEV system is the Quadratic Bidi-
rectional Buck-Boost Converter (QBBC), which efficiently
manages voltage requirements across the energy sources
and the electric motor. The control strategy for the motor
drive system employs an advanced Model Predictive Con-
trol (MPC) approach enhanced with a Recurrent Neural
Network (RNN). This RNN-based MPC system optimizes
motor performance and ensures high precision in speed
and torque control. The RNN model provides accurate pre-
dictions of motor speed, improving the responsiveness and
efficiency of the control system. To validate the effective-
ness of the proposed HEV system, a comprehensive simu-
lation study is conducted using MATLAB/SIMULINK. The
performance of the RNN-based MPC is benchmarked
against traditional control strategies, such as Artificial Neu-
ral Network-based MPC (ANN-MPC) and Proportional-Inte-
gral (PI) controllers. Additionally, the incremental conduct-
ance algorithm is employed as a Maximum Power Point
Tracking (MPPT) technique, ensuring optimal power extrac-
tion from the PV array. Simulation results demonstrate the
superior performance of the RNN-based MPC system, high-
lighting significant advancements in efficiency, precision,
and adaptability compared to conventional methods. This
research contributes to the advancement of intelligent and
sustainable HEV systems, emphasizing the potential of in-
tegrating renewable energy and advanced control tech-
niques to meet the evolving demands of modern transpor-
tation.
II. PROPOSED ELECTRIC VEHICLE WITH SOLAR ONBOARD
Figure 1 illustrates the proposed Hybrid Electric Vehicle
(HEV) design, which integrates onboard solar panels and
fuel cells. HEVs make use of specialized DC-DC bidirectional
buck-boost converters to enable power transfer in both di-
rections. In recent years, there has been growing interest
in high-gain boost converters, which are critical in applica-
tions such as fuel cells, distributed solar energy generation,
and backup power systems. High gain is defined by the ratio
of output voltage to input voltage (
𝑉𝑜𝑢𝑡
𝑉𝑖𝑛
). However, achiev-
ing significant step-up gains with traditional boost convert-
ers poses challenges, including the need for fast switching
times, which can lead to high voltage stress on components
and lower efficiency. Consequently, there is an increasing
demand for converters that provide higher power gains
while maintaining efficient performance. To achieve
greater output voltages, adjustments can be made to the
duty cycle or by increasing the number of windings in the
19
Відновлювана енергетика. №4/2024 | Комплексні проблеми енергетичних систем на основі НВДЕ
inductors. Nonetheless, converters using these techniques
often face issues, such as significant instabilities in input
current. An alternative approach is offered by quadratic
converters, which have gained considerable attention in
the power electronics community for their ability to en-
hance the output power of conventional boost converters.
Fig 1. Proposed HEV configuration
Fig 2. HEV with Solar and Fuel Cell
А. Electrolyzer
An electrolyzer utilizes electrical energy to decompose wa-
ter (H₂O) into hydrogen (H₂) and oxygen (O₂) through elec-
trolysis. This technique involves passing an electric current
through water or a saline solution, which splits the water
molecules into hydrogen and oxygen gases. These gases are
then collected separately and can be used in a range of ap-
plications. Hydrogen, for example, serves as a clean and ef-
ficient fuel for transportation, can be used in various indus-
trial processes, or powers fuel cells to generate electricity.
Electrolyzers typically operate at temperatures below 70°C
to prevent complications related to material degradation
and can achieve hydrogen purity levels of up to 99.9%.
The electrolytic process is driven by the flow of electric cur-
rent, denoted as 𝐼𝑒 . The amount of hydrogen produced can
be expressed as:
𝑋𝐻2
=
5.18𝑒−6𝐼𝑒𝑚𝑜𝑙𝑒
𝑠
(1)
𝑋𝐻2
denotes the molar flow rate of hydrogen, typically
given in units of moles per second 𝑚𝑜𝑙𝑒/𝑠. 𝐼𝑒 represents
the electric current passing through the electrochemical
system. The generated hydrogen is captured and stored in
a storage tank, where it serves as an additional energy
source. During periods when solar energy is insufficient, the
stored hydrogen is supplied to a fuel cell to generate elec-
tricity, ensuring a continuous energy supply.
В. Fuel cell
A fuel cell is a device that converts the energy stored in a
fuel, usually hydrogen, directly into electricity through a
chemical reaction. Unlike batteries, which store energy in-
ternally, fuel cells require a continuous supply of both fuel
and an oxidizing agent, such as oxygen from the air, to
maintain the energy conversion process. Proton Exchange
Membrane Fuel Cells (PEMFCs) are among the most used
fuel cell types. They operate at relatively low temperatures,
typically ranging from 50°C to 100°C, making them suitable
20
Відновлювана енергетика. №4/2024 | Комплексні проблеми енергетичних систем на основі НВДЕ
for a variety of applications, such as vehicle propulsion, sta-
tionary power systems for buildings, and portable elec-
tronic devices.
Output voltage from the fuel cell can be defined using
Nernst equation [24].
𝑉𝑜 = 𝐸𝑜 +
𝑅. 𝑇
2𝐹
ln
𝑥𝐻2
𝑥𝑂2
𝑥𝐻2𝑂
0.5
(2)
𝑉𝑜 is the cell potential under non-standard conditions (in
volts). 𝐸𝑜 is the standard cell potential (in volts), measured
under standard conditions (1 M concentration for solu-
tions, 1 atm pressure for gases, and pure solids or liquids
for other components). 𝑹 is the universal gas constant,
equal to 8.314 𝐽/(𝑚𝑜𝑙 · 𝐾). 𝑻 is the temperature in Kelvin.
𝑭 is Faraday's constant, approximately 96,485 𝐶/𝑚𝑜𝑙 (the
charge of one mole of electrons). 𝑥𝐻2
, 𝑥𝑂2
and 𝑥𝐻2𝑂 the
mole fractions (or partial pressures, assuming ideal gas be-
haviour) of hydrogen, oxygen, and water vapor, respec-
tively. A detailed analysis of the PV generation system, in-
corporating an incremental conductance-based MPPT
algorithm, along with a fuel cell integrated with a DC-DC
converter, is presented in [25].
С. Quadratic bidirectional boost/buck DC-DC converter
(QBBC)
Figure 4 presents the architecture of the Quadratic Bidirec-
tional Buck-Boost Converter (QBBC), which sets itself apart
from traditional boost quadratic converters by omitting ad-
ditional passive components like inductors and capacitors.
This innovative design streamlines the converter layout, re-
sulting in a more compact and efficient system without sac-
rificing performance. The QBBC regulates voltage through
a quadratic function, functioning effectively in both Boost
and Buck modes. A significant advantage of this converter
is its fixed voltage gain, which remains constant regardless
of the operational mode. The entire conversion process is
controlled by a single switch that oversees both charging
and discharging operations.
In its advanced operational mode, the QBBC acts as a step-
up converter, raising the voltage from the input to the out-
put. During this phase, switches 𝑇1 and 𝑇4 remain continu-
ously OFF, while the IGBT switch 𝑇3 is kept ON. This ar-
rangement facilitates efficient energy transfer at elevated
voltage levels. The output voltage regulation and inner cur-
rent loop control are achieved through PWM, where the
switching duration of 𝑇𝐶 is linked to switch 𝑇2. In boost
mode, when IGBT 𝑇2 is activated during the interval 𝛥𝑇𝑆,
inductors 𝐿1 and 𝐿2 are charged, resulting in a gradual in-
crease in current. This process allows energy stored in ca-
pacitor 𝐶 to be transferred to inductor 𝐿2.
Fig 3. Proposed HEV configuration with fuel cell and on-board PV panels
When the switch 𝑇2 is deactivated, the inductors 𝐿1 and 𝐿2
begin to discharge, resulting in a linearly decreasing cur-
rent. During this discharge phase, the energy stored in the
inductors is released to both the capacitor and the load, fa-
cilitating efficient power distribution. This discharge cycle is
essential for maintaining the desired output voltage and
ensuring that the converter functions effectively under var-
ying load conditions. The precise coordination of switching
states and the transfer of energy among the inductors, ca-
pacitors, and load enable the QBBC to operate efficiently in
both step-up and step-down modes. This design ensures re-
liable voltage regulation, making it suitable for a wide range
of applications.
∆𝑉𝑖𝑛 + (1 − Δ)(𝑉𝑖𝑛 − 𝑉𝑐) = 0 (3)
Δ𝑉𝑐 + (1 − Δ)(𝑉𝑐 − 𝑉𝑜𝑢𝑡) = 0 (4)
Then voltage gain is given as
𝐺 =
𝑉𝑜𝑢𝑡
𝑉𝑖𝑛
=
1
(1 − Δ)2
(5)
𝑉𝑖𝑛 input voltage of QBBC, 𝑉𝑐 capacitor voltage, 𝑉𝑜𝑢𝑡 output
voltage of QBBC and 𝐺 is the voltage gain.
In reverse mode, the Quadratic Bidirectional Buck-Boost
Converter (QBBC) operates as a buck converter, enabling
energy transfer from the output back to the input side. This
mode is essential for applications that require energy to be
redirected to the input, such as energy recovery or regen-
erative braking in electric vehicles.
21
Відновлювана енергетика. №4/2024 | Комплексні проблеми енергетичних систем на основі НВДЕ
Fig 4. QBBC schematic
Figure 5 depicts the control strategy utilized in the Quad-
ratic Bidirectional Buck-Boost Converter (QBBC), designed
for efficient operation in both boost and buck modes. This
adaptability is determined by the direction of power flow
within the system, enabling the converter to increase
(boost mode) or decrease (buck mode) the voltage as re-
quired. The operation of the DC-DC converter is regulated
by two main control loops: one for voltage regulation and
the other for current control. These loops are vital for main-
taining system stability and ensuring the converter func-
tions within its specified parameters. The voltage control
loop is tasked with ensuring a steady output voltage, even
when faced with changes in input voltage or load condi-
tions. It continuously monitors the output voltage and
compares it to a predetermined reference value. When dis-
crepancies arise, the loop modifies the duty cycle of the
switching components to rectify the output voltage. Em-
ploying a Proportional-Integral (PI) controller within this
loop helps minimize voltage errors over time, resulting in a
stable and precise output. Conversely, the current control
loop oversees the current passing through the converter,
ensuring it stays within safe operational limits. This loop is
crucial for safeguarding the converter's components from
overcurrent situations that could cause damage or de-
crease efficiency. Similar to the voltage control loop, the
current control loop also employs a PI controller to dynam-
ically adjust the converter's operations in response to any
variations between the actual and desired current levels.
Fig 5. QBBC control strategy
ІІІ. VECTOR CONTROL OF INDUCTION MOTOR
Figure 6 shows the indirect vector control by RNN speed es-
timate and MPC controller. Detailed explanation of vector
control of induction motor is presented in [26].
A. Modelling of RNN Observer
Figure 7 presents a detailed block diagram of the proposed
model system, incorporating an advanced observer de-
signed to estimate the rotor flux and rotor speed of an in-
duction motor. This observer is integral to the system, uti-
lizing only stator voltage and stator current as input signals,
which enhances its efficiency for real-time monitoring and
control.
The observer is developed using a Recurrent Neural Net-
work (RNN), well-suited for processing time-series data and
managing dynamic systems. It comprises two primary com-
ponents: the Rotor Flux Observer and the Rotor Speed Ob-
server. The Rotor Flux Observer estimates the rotor flux
based on the input signals from the stator voltage and cur-
rent. This estimation's accuracy is vital, as it significantly im-
pacts the performance of the rotor speed observer. In turn,
the Rotor Speed Observer takes the stator current and the
22
Відновлювана енергетика. №4/2024 | Комплексні проблеми енергетичних систем на основі НВДЕ
estimated rotor flux from the Rotor Flux Observer to pro-
vide an accurate estimation of the rotor speed, which is cru-
cial for effective motor control.
The RNN-based observer is based on the dynamic equa-
tions that govern the operation of the induction motor. By
leveraging these equations, the observer can effectively
monitor and predict the motor's behavior under various
operating scenarios. The RNN is trained using a dataset
generated from the induction motor model, which includes
essential inputs such as stator voltage, stator current, and
rotor speed. This dataset serves as the foundation for the
RNN to learn and build an accurate estimation model for
rotor flux and speed.
Fig 6. Implementation of the indirect vector control
Additionally, Figure 7 depicts the learning process of the
RNN observer. During this phase, the RNN is trained with
historical data to enhance its predictive accuracy for rotor
flux and speed. This training is crucial to ensure the ob-
server can reliably estimate these parameters in real-time
applications.
After the training is complete, the RNN observer's configu-
ration is established to guarantee precise and dependable
performance. The Rotor Flux Observer accepts the stator
voltage and current as inputs, generating an estimate of the
rotor flux, which is essential for the subsequent speed esti-
mation. Meanwhile, the Rotor Speed Observer uses the es-
timated rotor flux from the Rotor Flux Observer along with
the stator current to estimate the rotor speed. This ar-
rangement promotes a robust estimation process, facilitat-
ing accurate control and monitoring of the induction mo-
tor. The design of the Recurrent Neural Network (RNN)
architecture is a vital phase in developing the observer. This
design involves selecting essential parameters, including
the number of hidden layers and the quantity of neurons
within each layer, which significantly influence the net-
work's learning capabilities. The complexity of the network
is tailored according to the learning error observed during
training. If the learning error surpasses the desired thresh-
old, additional hidden layers are incorporated to boost the
network’s ability to learn.
In this study, the RNN architecture is configured to include
20 neurons in each hidden layer, achieving a balance be-
tween complexity and learning efficiency. The learning rate
is set at 0.4, which dictates how quickly the network's
weights are adjusted. A target maximum allowable error of
0.005 is established for the learning process, ensuring that
the network attains a high level of accuracy. The training is
limited to a maximum of 500 epochs, allowing ample itera-
tions for the network to converge toward an optimal solu-
tion. The training employs the back-propagation algorithm,
which updates the network's weights iteratively based on
the discrepancies between the predicted outputs and the
actual values. This process gradually decreases the learning
error and enhances the observer's accuracy.
Fig 7. RNN based Speed observer
23
Відновлювана енергетика. №4/2024 | Комплексні проблеми енергетичних систем на основі НВДЕ
IV. SIMULATION RESULTS
The proposed Hybrid Electric Vehicle (HEV) system repre-
sents a sophisticated configuration that incorporates an
onboard photovoltaic (PV) array and is governed by an ad-
vanced Recurrent Neural Network-based Model Predictive
Control (RNN-MPC) framework. This innovative system is
rigorously simulated using MATLAB/SIMULINK to assess its
performance and adaptability across various operational
scenarios. In this cutting-edge HEV design, the fuel cell acts
as the primary energy source for the vehicle, supplying the
necessary power to drive the induction motor that propels
the vehicle. Complementing the fuel cell is an onboard PV
array, which captures solar energy to provide supplemen-
tary power, particularly during sunny conditions. This inte-
gration not only alleviates the burden on the fuel cell but
also enhances the overall efficiency of the vehicle.
A notable feature of this HEV system is its capability for re-
generative braking, which allows it to recover energy during
braking or deceleration of the induction motor. Rather than
losing this kinetic energy as heat, the system efficiently cap-
tures it and directs it to a hydrogen storage tank. The conver-
sion of this electrical energy into chemical energy is facili-
tated by an electrolyzer, which splits water molecules into
hydrogen and oxygen, thus storing the hydrogen for later
use. This stored hydrogen can be utilized either by the
onboard fuel cell or for other applications, further improving
the vehicle’s energy efficiency and extending its operational
range. To control the speed of the induction motor, the sys-
tem utilizes a sensorless indirect vector control technique.
This effective method regulates the motor’s speed and
torque without requiring direct measurements of the rotor
position. Instead, it estimates the rotor flux and position in-
directly by analysing the stator current and voltage, subse-
quently adjusting the motor control signals as needed.
This vector control strategy is integrated with a Model Pre-
dictive Control (MPC) system tailored for speed regulation.
Enhanced by a Recurrent Neural Network (RNN), this MPC
framework acts as an estimator for the induction motor’s
speed. The RNN is trained on historical operational data,
enabling it to accurately predict the motor’s speed in real-
time. This predictive functionality ensures that the motor
operates efficiently and can quickly adapt to changes in
load or driving conditions. The complete HEV system is
modelled and simulated in MATLAB/SIMULINK to evaluate
its performance across various conditions. The simulation
encompasses different scenarios, including:
• Variations in Solar Irradiation: The system's reaction to
fluctuations in sunlight intensity is examined to assess the
efficiency of the PV array in delivering additional energy.
• Changes in Vehicle Velocity: The simulation investi-
gates how the system manages the vehicle's speed and
energy consumption under different driving speeds.
• Fluctuations in Load and Force: The system's capability
to respond to varying load conditions, such as driving
uphill or transporting extra weight, is assessed.
These simulations yield critical insights into the system's
ability to sustain stable and efficient operation in diverse
real-world scenarios. The outcomes contribute to under-
standing the effectiveness of the proposed HEV system,
particularly regarding its energy management strategies,
precision in speed control, and overall adaptability.
Case 1
In this simulation scenario, the effectiveness and efficiency
of the proposed HEV system are evaluated under controlled
conditions. The photovoltaic (PV) arrays operate at a con-
stant irradiance level to maintain a stable energy input, while
the vehicle's speed is varied to represent different driving
scenarios. The irradiance level is fixed at 1000 W/m²
throughout the simulation, which lasts from 0 to 20 seconds.
This setup guarantees that the PV arrays produce a con-
sistent power output, allowing for a focused analysis on the
impact of the vehicle's speed on overall system performance.
During this period, the vehicle's reference speed is adjusted
to test the system's adaptability:
• From 0 to 10 seconds, the reference speed is set at 120
rad/s.
• From 10 to 20 seconds, the reference speed decreases
to 140 rad/s.
These speed variations simulate real-world driving situa-
tions where the vehicle may need to accelerate or deceler-
ate based on road and traffic conditions.
Figure 8 illustrates a comparison of different control strat-
egies—PI control, ANN-based Model Predictive Control
(MPC), and RNN-based MPC—in regulating the vehicle's
speed and torque. The results indicate that RNN-based
MPC significantly improves vehicle performance over tradi-
tional PI control, reducing peak overshoot, rise time, and
settling time for a smoother and more stable operation. Alt-
hough ANN-based MPC also outperforms PI control, it does
not achieve the same level of precision and stability as the
RNN-based approach. This comparison emphasizes the su-
perior effectiveness of RNN-based MPC in delivering re-
fined control, enhancing vehicle handling and perfor-
mance. Figure 9 presents the voltage, current, and power
output from the solar panels during the constant irradiance
phase. An Incremental Conductance algorithm is employed
for maximum power point tracking (MPPT) to optimize en-
ergy generation from the PV array. This algorithm is vital for
adjusting the operating point of the PV array to ensure
maximum efficiency under constant irradiance conditions.
At 1000 W/m², the solar panels generate a power output of
2.01 kilowatts, meeting the vehicle's energy demands
throughout the simulation and ensuring the PV array's ef-
fective contribution to vehicle operation.
Figures 10 through 14 provide an in-depth look at the energy
management within the HEV system. Figure 10 displays the
power, voltage, and current generated by the PV converter,
which adapts the harvested energy from the PV array for use
by the vehicle's electrical systems. Figures 11 and 12 show-
case the power, voltage, and current from the fuel cell (FC)
and its corresponding converter. The fuel cell serves as the
primary power source, with the converter ensuring the en-
ergy is appropriately regulated for the vehicle's needs. Fig-
ures 13 and 14 illustrate the power, voltage, and current re-
lated to the electrolyzer, which is crucial for converting
surplus electrical energy into chemical energy, stored as hy-
drogen for later use. Figure 15 highlights the performance of
24
Відновлювана енергетика. №4/2024 | Комплексні проблеми енергетичних систем на основі НВДЕ
the Quadratic Bidirectional Buck-Boost Converter (QBBC),
which is critical for increasing the input voltage from 400V to
780V. This voltage boost is essential for powering the vehi-
cle's drivetrain and other high-voltage components. The
QBBC's efficiency in stepping up voltage is vital for maintain-
ing vehicle performance, especially during periods of high-
power demand. Total Harmonic Distortion (THD) in the sta-
tor current is a key parameter for assessing the quality of
electrical power delivered to the motor; lower THD values
signify smoother power delivery, which minimizes wear on
motor components and enhances overall efficiency.
Fig 8. Motor speed and torque with PI control, ANN-MPC
control and with proposed RNN-MPC control
Fig 9. PV output voltage, current and power
With PI control, the THD is measured at 5.3%, indicating a
relatively high level that could lead to inefficiencies and po-
tential long-term motor damage. The implementation of
ANN-based MPC improves the THD to 2.8%, marking a sig-
nificant enhancement in power quality. The proposed RNN-
based MPC further reduces the THD to 1.2%, demonstrat-
ing the best performance among the control strategies
tested. This low THD signifies that the RNN-based MPC ef-
fectively delivers clean and stable power to the motor, en-
suring optimal operation and longevity.
Fig 10. PV converter output voltage, current and power
Fig 11. Fuel vell voltage and current
25
Відновлювана енергетика. №4/2024 | Комплексні проблеми енергетичних систем на основі НВДЕ
Fig 12. Fuel cell converter voltage and current
Fig 13. Electrolyzer output voltage and current
The simulation aims to assess the performance of a future
electric vehicle (EV) configuration by modeling varying sun-
light conditions alongside different vehicle speeds. This
thorough analysis provides a detailed understanding of the
EV system's behaviour under realistic and dynamic operat-
ing conditions, with a specific focus on the effectiveness of
various control strategies.
The simulation encompasses a series of speed variations
that mimic different driving scenarios, including:
0 seconds: Speed set at 150 rad/s.
2.5 seconds: Speed increases to 120 rad/s.
5 seconds: Further increases to 100 rad/s.
7.5 seconds: Speed ramps up to 80 rad/s.
10 seconds: Accelerates to 60 rad/s.
12.5 seconds: Speed reaches 40 rad/s.
15 seconds: Peaks at 20 rad/s.
17.5 seconds: Decelerates back to 100 rad/s.
Fig 14. Electrolyzer converter output voltage and current
Fig 15. QBBC input voltage and output voltage
Case 2:
At the same time, the solar irradiance, which directly influ-
ences the power output from the PV arrays, is varied to re-
flect changing sunlight conditions:
26
Відновлювана енергетика. №4/2024 | Комплексні проблеми енергетичних систем на основі НВДЕ
0 seconds: Irradiance starts at 500 W/m².
3 seconds: Increases to 700 W/m².
6 seconds: Peaks at 900 W/m².
9 seconds: Slightly decreases to 800 W/m².
12 seconds: Further reduces to 600 W/m².
15 seconds: Drops to 400 W/m².
18 seconds: Reaches a low of 200 W/m².
Figure 16 provides a comparative overview of the EV's mo-
tor speed and torque under three control strategies: PI con-
trol, ANN-based Model Predictive Control (MPC), and the
proposed RNN-based MPC. This figure is essential for ana-
lysing how each control method manages the dynamic var-
iations in speed and irradiance. As a traditional control
method, PI control demonstrates adequate performance
but struggles with rapid changes in speed and irradiance,
resulting in higher overshoot, longer settling times, and de-
creased stability. ANN-based MPC shows improvements
over PI control by offering enhanced adaptability and re-
ducing overshoot magnitude, yet it still encounters chal-
lenges under extreme conditions. In contrast, RNN-based
MPC significantly surpasses both PI and ANN-based MPC,
delivering smoother transitions, minimal overshoot,
quicker settling times, and greater stability during rapid
changes in vehicle speed and fluctuating irradiance levels.
Figure 17 displays the output power, voltage, and current
from the solar panels under varying irradiance conditions.
The results highlight the PV system's responsiveness to
changing sunlight, with the Maximum Power Point Tracking
(MPPT) algorithm optimizing energy extraction at each irra-
diance level. The MPPT ensures that the solar panels oper-
ate at their optimal point, providing consistent power to
the EV system even amidst fluctuating light conditions. Fig-
ure 18 shows the output voltage, current, and power pro-
duced by the fuel cell and its converter, which serves as the
primary power source, particularly during periods of insuf-
ficient solar energy. This ensures that the EV has a depend-
able power supply even in low irradiance scenarios, main-
taining vehicle operation.
Figure 19 focuses on the output power, voltage, and cur-
rent associated with the electrolyzer converter. The elec-
trolyzer efficiently converts surplus electrical energy into
chemical energy stored as hydrogen, which can later be uti-
lized to power the vehicle or other systems. This capability
highlights the system's effectiveness in managing energy
and storing excess power for future use. Figure 20 under-
scores the importance of the Quadratic Bidirectional Buck-
Boost Converter (QBBC), which regulates and boosts the
output voltage, especially during fluctuating solar irradi-
ance. The QBBC increases the input voltage from the PV ar-
rays to the higher level required by the vehicle's drive in-
verter, ensuring reliable and consistent power delivery.
This voltage regulation is critical for maintaining vehicle
performance and preventing voltage drops that could neg-
atively impact the operation of the drive inverter.
This comprehensive simulation study emphasizes the signifi-
cance of advanced control strategies and robust energy man-
agement systems in contemporary EV configurations. The pro-
posed RNN-based MPC showcases superior performance in
managing vehicle speed and torque under varying conditions,
providing enhanced stability, reduced overshoot, and faster
response times compared to traditional PI and ANN-based
MPC methods. Furthermore, the effective management of en-
ergy from the solar panels, fuel cells, and electrolyzer, coupled
with the reliable voltage regulation afforded by the QBBC, en-
sures that the EV system operates efficiently and effectively,
even in variable environmental conditions. This thorough anal-
ysis not only demonstrates the advantages of the proposed
system but also offers valuable insights into its potential appli-
cations in future EV technologies.
Fig 16. Motor speed and torque with PI control, ANN-MPC
control and with Pproposed RNN-MPC control
Fig 17. PV converter output voltage, current and power
27
Відновлювана енергетика. №4/2024 | Комплексні проблеми енергетичних систем на основі НВДЕ
Fig 18. Fuel cell converter output voltage, current and power
Fig 19. Electrolyzer converter output voltage, current and power
Fig 20. QBBC input voltage and output voltage
V. CONCLUSION
In this study, we have presented an innovative Hybrid Elec-
tric Vehicle (HEV) configuration that integrates an onboard
photovoltaic (PV) array and employs an advanced Recur-
rent Neural Network-based Model Predictive Control (RNN-
MPC) system. The comprehensive simulations conducted
using MATLAB/SIMULINK evaluated the performance and
adaptability of the proposed system under a variety of op-
erational conditions, demonstrating its effectiveness in
managing energy resources and maintaining vehicle perfor-
mance. The simulation results reveal that the RNN-based
MPC significantly outperforms traditional control methods,
including Proportional-Integral (PI) control and Artificial
Neural Network (ANN)-based MPC, in terms of speed regu-
lation and torque management. The RNN-based approach
not only provides smoother transitions and reduced over-
shoot but also enhances stability and response times during
rapid changes in vehicle speed and fluctuating solar irradi-
ance levels. This superior performance underscores the ad-
vantages of utilizing RNNs for real-time predictions and
control in dynamic environments. Furthermore, the sys-
tem's ability to effectively manage energy from the PV ar-
ray, fuel cell, and electrolyzer highlights its potential for ef-
ficient energy utilization and storage. The integration of
regenerative braking and hydrogen production through the
electrolyzer demonstrates a holistic approach to energy
management, extending the operational range of the vehi-
cle while minimizing dependency on conventional fuel
sources. The findings of this study provide valuable insights
into the future of HEV systems, emphasizing the im-
portance of advanced control strategies and robust energy
management frameworks. The proposed configuration not
only showcases the potential for improved vehicle perfor-
mance but also contributes to the ongoing efforts in devel-
oping sustainable and environmentally friendly transporta-
tion solutions. Future research may focus on optimizing the
RNN architecture further and exploring the integration of
additional renewable energy sources, paving the way for
the next generation of intelligent and eco-efficient electric
vehicles.
REFERENCES
1. Shah, K. J., Pan, S. Y., Lee, I., Kim, H., You, Z., Zheng, J. M.,
& Chiang, P. C. (2021). Green transportation for sustain-
ability: Review of current barriers, strategies, and innova-
tive technologies. Journal of Cleaner Production, 326,
129392.
2. Waseem, M., Amir, M., Lakshmi, G. S., Harivardhagini,
S., & Ahmad, M. (2023). Fuel cell-based hybrid electric
vehicles: An integrated review of current status, key
challenges, recommended policies, and future pro-
spects. Green Energy and Intelligent Transportation,
100121.
3. Hassan, I. A., Ramadan, H. S., Saleh, M. A., & Hissel, D.
(2021). Hydrogen storage technologies for stationary and
28
Відновлювана енергетика. №4/2024 | Комплексні проблеми енергетичних систем на основі НВДЕ
mobile applications: Review, analysis and perspectives. Re-
newable and Sustainable Energy Reviews, 149, 111311.
4. Xu, Q., Vafamand, N., Chen, L., Dragičević, T., Xie, L., &
Blaabjerg, F. (2020). Review on advanced control tech-
nologies for bidirectional DC/DC converters in DC mi-
crogrids. IEEE Journal of Emerging and Selected Topics
in Power Electronics, 9(2), 1205-1221.
5. Maheswari, E., Kommula, B. N., Rajesh, K., & Sharma, S.
(2024). Optimized maximum power point tracking for
PV fed hybrid DC-DC converter-based BLDC motor
driven electric vehicles. Electrical Engineering, 1-22.
6. Naqvi, S. S. A., Jamil, H., Iqbal, N., Khan, S., Lee, D. I.,
Park, Y. C., & Kim, D. H. (2024). Multi-objective optimi-
zation of PI controller for BLDC motor speed control and
energy saving in Electric Vehicles: A constrained swarm-
based approach. Energy Reports, 12, 402-417.
7. Frindi, K. Artificial neural network control design for elec-
tric vehicle fast charging station using Vienna rectifier.
8. Luo, Y., Wu, Y., Li, B., Qu, J., Feng, S. P., & Chu, P. K.
(2021). Optimization and cutting-edge design of fuel-
cell hybrid electric vehicles. International Journal of En-
ergy Research, 45(13), 18392-18423.
9. Pathak, P. K., Padmanaban, S., Yadav, A. K., Alvi, P. A., &
Khan, B. (2022). Modified incremental conductance
MPPT algorithm for SPV-based grid-tied and stand-
alone systems. IET Generation, Transmission & Distribu-
tion, 16(4), 776-791.
10. Hassan, M. M., Shaikh, M. S., Jadoon, H. U. K., Atif, M. R.,
& Sardar, M. U. (2020). Dynamic modeling and vector con-
trol of AC induction traction motor in china railway. Sukkur
IBA Journal of Emerging Technologies, 3(2), 115-125.
11. Guerrero, J., Gebbran, D., Mhanna, S., Chapman, A. C.,
& Verbič, G. (2020). Towards a transactive energy sys-
tem for integration of distributed energy resources:
Home energy management, distributed optimal power
flow, and peer-to-peer energy trading. Renewable and
Sustainable Energy Reviews, 132, 110000.
12. Zhao, L., Yao, W., Wang, Y., & Hu, J. (2020). Machine
learning-based method for remaining range prediction
of electric vehicles. Ieee Access, 8, 212423-212441.
13. Keshavarzi, M. D., & Ali, M. H. (2020). A novel bidirec-
tional dc-dc converter for dynamic performance en-
hancement of hybrid AC/DC microgrid. Electronics,
9(10), 1653.
14. Rezaei, H., Abdollahi, S. E., Abdollahi, S., & Filizadeh, S.
(2022). Energy management strategies of battery-ultra-
capacitor hybrid storage systems for electric vehicles:
Review, challenges, and future trends. Journal of Energy
Storage, 53, 105045.
15. Millo, F., Rolando, L., Tresca, L., & Pulvirenti, L. (2023).
Development of a neural network-based energy mana-
gement system for a plug-in hybrid electric vehicle.
Transportation Engineering, 11, 100156.
16. Awad, M., Ibrahim, A. M., Alaas, Z. M., El-Shahat, A., &
Omar, A. I. (2022). Design and analysis of an efficient
photovoltaic energy-powered electric vehicle charging
station using perturb and observe MPPT algorithm.
frontiers in energy research, 10, 969482.
17. Tolj, I., Penga, Ž., Vukičević, D., & Barbir, F. (2020). Ther-
mal management of edge-cooled 1 kW portable proton
exchange membrane fuel cell stack. Applied Energy,
257, 114038.
18. Zhang, F., Hu, X., Langari, R., Wang, L., Cui, Y., & Pang,
H. (2021). Adaptive energy management in automated
hybrid electric vehicles with flexible torque request. En-
ergy, 214, 118873.
19. Ponnapalli, B., Lakshmikhandan, K., Palanisamy, K., &
Sithambaram, M. (2024). A hybrid technique for grid-
connected solar–wind hybrid system with electric vehi-
cles. Energy & Environment, 35(5), 2753-2789.
20. Babu, T. S., Vasudevan, K. R., Ramachandaramurthy, V.
K., Sani, S. B., Chemud, S., & Lajim, R. M. (2020). A com-
prehensive review of hybrid energy storage systems:
Converter topologies, control strategies and future pro-
spects. IEEE Access, 8, 148702-148721.
21. Farhadi Gharibeh, H., & Farrokhifar, M. (2021). Online
multi-level energy management strategy based on rule-
based and optimization-based approaches for fuel cell
hybrid electric vehicles. Applied Sciences, 11(9), 3849.
22. Mousavi, M. S., Davari, S. A., Nekoukar, V., Garcia, C.,
He, L., Wang, F., & Rodriguez, J. (2022). Predictive
torque control of induction motor based on a robust in-
tegral sliding mode observer. IEEE Transactions on In-
dustrial Electronics, 70(3), 2339-2350.
23. Lan, T., Jermsittiparsert, K., T. Alrashood, S., Rezaei, M.,
Al-Ghussain, L., & A. Mohamed, M. (2021). An advanced
machine learning based energy management of renew-
able microgrids considering hybrid electric vehicles’
charging demand. Energies, 14(3), 569.
24. Sterratt, D. C. (2022). Nernst Equation. In Encyclopedia
of Computational Neuroscience (pp. 2181-2182). New
York, NY: Springer New York.
25. Divya, G., & Padmavathi, S. V. (2024). Intelligent Power
Management of Electric vehicle with onboard PV by
ANN-based Model Predictive Control. Journal of Electri-
cal Systems, 20(2), 2395-2417.
26. Boulmane, A., Zidani, Y., Chennani, M., & Belkhayat, D.
(2020). Design of robust adaptive observer against
measurement noise for sensorless vector control of in-
duction motor drives. Journal of Electrical and Com-
puter Engineering, 2020(1), 6570942.
|
| id | veorgua-article-483 |
| institution | Vidnovluvana energetika |
| keywords_txt_mv | keywords |
| language | English |
| last_indexed | 2026-07-19T01:14:16Z |
| publishDate | 2024 |
| publisher | Institute of Renewable Energy National Academy of Sciences of Ukraine |
| record_format | ojs |
| resource_txt_mv | veorgua/54/f7cbebe3ae38fd6bb39040502d2f8754.pdf |
| spelling | veorgua-article-4832026-07-18T06:32:20Z RNN-INTEGRATED MODEL PREDICTIVE CONTROL FOR FUEL CELL AND SOLAR-POWERED HYBRID ELECTRIC VEHICLES RNN-ПОКРАЩЕНА МОДЕЛЬ ПРОГНОЗОВАНОГО КЕРУВАННЯ ДЛЯ ГІБРИДНИХ ЕЛЕКТРИЧНИХ ТРАНСПОРТНИХ ЗАСОБІВ З ПАЛИВНИМИ ЕЛЕМЕНТАМИ ТА ФОТОЕЛЕКТРИЧНИМИ ДЖЕРЕЛАМИ ЕНЕРГІЇ Divya, G. Venkata Padmavathi , S. Hybrid Electric Vehicle (HEV), Recurrent Neural Network (RNN), Fuel Cell, Induction Motor Control, Electro-lyzer, Renewable Energy Integration, Speed Control, MATLAB/SIMULINK Simulation. Гібридний електричний транспортний засіб (HEV), рекурентна нейронна мережа (RNN), паливний елемент, управління індукційним двигуном, електролізер, інтеграція відновлюваних джерел енергії, контроль швидкості, симуляція в MATLAB/SIMULINK. This paper presents an innovative Hybrid Electric Vehicle (HEV) configuration utilizing a fuel cell as the primary energy source and an onboard Photovoltaic (PV) array as a supplementary source. The system features an advanced Model Predictive Control (MPC) enhanced by a Recurrent Neural Network (RNN) to manage the induction motor efficiently. Key components include a PV array, a fuel cell, and an electrolyzer. The PV array supplements the fuel cell during optimal sunlight conditions, while excess energy during idle periods is converted to hydrogen via the electrolyzer and stored in a hydrogen tank for future use. A quadratic bidirectional buck-boost converter (QBBC) regulates voltage, ensuring compatibility between energy sources and the motor. The system’s performance is evaluated under various sunlight and speed conditions, with the RNN-based MPC compared to an Artificial Neural Network-based MPC (ANN-MPC) and a traditional Proportional-Integral (PI) controller. An incremental conductance algorithm is implemented for Maximum Power Point Tracking (MPPT) to optimize PV power extraction. The RNN model predicts motor speed, enhancing control precision. Simulations in MATLAB/SIMULINK reveal that the RNN-based MPC outperforms ANN-MPC and PI controllers, demonstrating improved efficiency and speed control. This work contributes to advancing intelligent and energy-efficient HEV technologies. Ця робота представляє інноваційну конфігурацію гібридного електричного транспортного засобу (HEV), що використовує паливний елемент як основне джерело енергії та вбудовану фотоелектричну (PV) панель як додаткове джерело. Система оснащена вдосконаленим прогнозуючим управлінням на основі моделей (Model Predictive Control, MPC), посиленим рекурентною нейронною мережею (RNN), для ефективного керування індукційним двигуном. Основні компоненти включають PV-панель, паливний елемент та електролізер. PV-панель доповнює роботу паливного елемента за оптимальних умов освітлення, а надлишкова енергія у періоди простою перетворюється на водень за допомогою електролізера та зберігається в водневому баку для подальшого використання. Квадратичний двонаправлений понижуючо-підвищуючий перетворювач (QBBC) регулює напругу, забезпечуючи сумісність між джерелами енергії та двигуном. Продуктивність системи оцінюється за різних умов освітлення та швидкості, а MPC на основі RNN порівнюється з MPC на основі штучної нейронної мережі (ANN-MPC) та традиційним пропорційно-інтегральним (PI) контролером. Для відстеження точки максимальної потужності (Maximum Power Point Tracking, MPPT) впроваджений алгоритм інкрементальної провідності, щоб оптимізувати витяг енергії з PV-панелі. Модель RNN прогнозує швидкість двигуна, підвищуючи точність управління. Симуляції в MATLAB/SIMULINK показують, що MPC на основі RNN перевершує ANN-MPC та PI-контролери, демонструючи кращу ефективність і контроль швидкості. Ця робота сприяє розвитку інтелектуальних та енергоефективних технологій HEV. Institute of Renewable Energy National Academy of Sciences of Ukraine 2024-12-10 Article Article application/pdf https://ve.org.ua/index.php/journal/article/view/483 10.36296/1819-8058.2024.4(79).17-28 Vidnovluvana energetika ; No. 4(79) (2024): Scientific and applied Journal renewable energy ; 17-28 Возобновляемая энергетика; ##issue.no## 4(79) (2024): Scientific and applied Journal renewable energy ; 17-28 Відновлювана енергетика; № 4(79) (2024): Науково-прикладний журнал Відновлювана енергетика; 17-28 2664-8172 1819-8058 10.36296/1819-8058.2024.4(79) en https://ve.org.ua/index.php/journal/article/view/483/392 Copyright (c) 2024 G. Divya, S. Venkata Padmavathi https://creativecommons.org/licenses/by-nc-nd/4.0 |
| spellingShingle | Hybrid Electric Vehicle (HEV) Recurrent Neural Network (RNN) Fuel Cell Induction Motor Control Electro-lyzer Renewable Energy Integration Speed Control MATLAB/SIMULINK Simulation. Divya, G. Venkata Padmavathi , S. RNN-INTEGRATED MODEL PREDICTIVE CONTROL FOR FUEL CELL AND SOLAR-POWERED HYBRID ELECTRIC VEHICLES |
| title | RNN-INTEGRATED MODEL PREDICTIVE CONTROL FOR FUEL CELL AND SOLAR-POWERED HYBRID ELECTRIC VEHICLES |
| title_alt | RNN-ПОКРАЩЕНА МОДЕЛЬ ПРОГНОЗОВАНОГО КЕРУВАННЯ ДЛЯ ГІБРИДНИХ ЕЛЕКТРИЧНИХ ТРАНСПОРТНИХ ЗАСОБІВ З ПАЛИВНИМИ ЕЛЕМЕНТАМИ ТА ФОТОЕЛЕКТРИЧНИМИ ДЖЕРЕЛАМИ ЕНЕРГІЇ |
| title_full | RNN-INTEGRATED MODEL PREDICTIVE CONTROL FOR FUEL CELL AND SOLAR-POWERED HYBRID ELECTRIC VEHICLES |
| title_fullStr | RNN-INTEGRATED MODEL PREDICTIVE CONTROL FOR FUEL CELL AND SOLAR-POWERED HYBRID ELECTRIC VEHICLES |
| title_full_unstemmed | RNN-INTEGRATED MODEL PREDICTIVE CONTROL FOR FUEL CELL AND SOLAR-POWERED HYBRID ELECTRIC VEHICLES |
| title_short | RNN-INTEGRATED MODEL PREDICTIVE CONTROL FOR FUEL CELL AND SOLAR-POWERED HYBRID ELECTRIC VEHICLES |
| title_sort | rnn-integrated model predictive control for fuel cell and solar-powered hybrid electric vehicles |
| topic | Hybrid Electric Vehicle (HEV) Recurrent Neural Network (RNN) Fuel Cell Induction Motor Control Electro-lyzer Renewable Energy Integration Speed Control MATLAB/SIMULINK Simulation. |
| topic_facet | Hybrid Electric Vehicle (HEV) Recurrent Neural Network (RNN) Fuel Cell Induction Motor Control Electro-lyzer Renewable Energy Integration Speed Control MATLAB/SIMULINK Simulation. Гібридний електричний транспортний засіб (HEV) рекурентна нейронна мережа (RNN) паливний елемент управління індукційним двигуном електролізер інтеграція відновлюваних джерел енергії контроль швидкості симуляція в MATLAB/SIMULINK. |
| url | https://ve.org.ua/index.php/journal/article/view/483 |
| work_keys_str_mv | AT divyag rnnintegratedmodelpredictivecontrolforfuelcellandsolarpoweredhybridelectricvehicles AT venkatapadmavathis rnnintegratedmodelpredictivecontrolforfuelcellandsolarpoweredhybridelectricvehicles AT divyag rnnpokraŝenamodelʹprognozovanogokeruvannâdlâgíbridnihelektričnihtransportnihzasobívzpalivnimielementamitafotoelektričnimidžerelamienergíí AT venkatapadmavathis rnnpokraŝenamodelʹprognozovanogokeruvannâdlâgíbridnihelektričnihtransportnihzasobívzpalivnimielementamitafotoelektričnimidžerelamienergíí |