E3S Web of Conferences 387, 04008 (2023) https://doi.org/10.
1051/e3sconf/202338704008
ICSERET-2023
A MACHINE LEARNING-BASED ENERGY
OPTIMIZATION SYSTEM FOR ELECTRIC
VEHICLES
Padmavathy R 1 Jeya Prakash K.2,.Greeta T 3,.Divya K 4
1
New Prince Shrishri Bhavani College Of Engineering And Technology,
Approved by AICTE, Affilated to Anna UniversityChennai, India
2
Assistant Professor Kalasalingam Academy of Research and Education
Tamilnadu, India
3
Assistant Professor, Prince Dr.K.Vasudevan College of Engineering and Technology, India
4
Assistant Professor, Prince Shri Venkateshwara Padmavathy Engineering College, India.
Abstract- The growing demand for sustainable and eco-friendly
transportation has led to the widespread adoption of electric vehicles
(EVs). However, the limited driving range of EVs and the need for
frequent recharging remain significant challenges. To address these
challenges, researchers have proposed various energy optimization
techniques, including machine learning-based approaches. In this paper,
proposed method of Smart EV energy optimization systems for EVs. The
system uses machine learning algorithms to analyze and learn from
historical driving data, such as the driving patterns, road conditions,
weather, and traffic. Based on this analysis, the system predicts the energy
consumption of the EV and optimizes the energy usage to minimize energy
waste and extend the driving range.
Keywords: Electric vehicles, Energy optimization, Machine learning,
Optimizationalgorithms, Range extension;
1. Introduction
Electric vehicles (EVs) have become increasingly popular in recent years due to their
environmental benefits and potential for reducing dependence on fossil fuels. However, one
major challenge facing EVs is limited driving range, which can be a significant barrier to
widespread adoption. [1] To address this issue, researchers and engineers have been
developing various energy optimization strategies and technologies to improve the
efficiency and extend the range of EVs.
To address these challenges, researchers have developed a machine learning-based
energy optimization system for EVs. [2] The system uses machine learning algorithms to
analyze and learn from historical driving data and predict energy consumption. Based on
the predictions, the system optimizes energy usage to minimize energy waste and extend
the driving range.
© The Authors, published by EDP Sciences. This is an open access article distributed under the terms of the Creative Commons Attribution
License 4.0 (https://creativecommons.org/licenses/by/4.0/).
E3S Web of Conferences 387, 04008 (2023) https://doi.org/10.1051/e3sconf/202338704008
ICSERET-2023
One promising approach is the use of machine learning algorithms to predict energy
consumption and optimize energy usage. [3] Machine learning has the potential to learn
from historical driving data, such as driving patterns, traffic, and weather conditions, and
provide personalized energy management solutions for individual drivers. [4] By
optimizing energy usage, the system can minimize energy waste and extend the driving
range of the EV.
In this paper, we present a machine learning-based energy optimization system for EVs,
which has been developed and evaluated through simulations and field tests. [5] The system
takes into account various factors such as battery condition, charging status, and available
charging infrastructure to provide personalized energy management solutions. The system's
performance has been compared with other energy optimization methods, demonstrating
superior results.
Energy Management
Strategy for Hybrid
Electric Vehicles
Optimization -
Rule-based
based Control
Control Strategy
Strategy
Deterministic Rule- Global
based Control Optimization
Strategy Control Strategy
Fuzzy Logic Rule- Instantaneous
based Control Optimization
Strategy Control Strategy
Figure 1. Classification of energy management for EV
This paper provides an overview of the machine learning-based energy optimization
system, including the system architecture, algorithm design, and performance evaluation.
The results show that the system can significantly improve energy efficiency and range
extension, making it a promising approach to enhance the performance and adoption of
EVs. The paper also discusses the potential implications and future directions of this
technology in the development of sustainable transportation systems.
2. Literature Review
1. Li [6] et al. (2022) proposed a machine learning approach for energy optimization of
EVs using driving pattern recognition. They used driving data from real-world EVs to train
their model and achieved high accuracy in predicting the energy demand of Support Vector
Machine-based Energy Optimization for Electric Mobility. However, their model was
limited to specific driving patterns and did not consider factors such as traffic conditions.
Merits
2
E3S Web of Conferences 387, 04008 (2023) https://doi.org/10.1051/e3sconf/202338704008
ICSERET-2023
1. The system can adapt to different driving patterns, allowing for more efficient
energy management over a range of driving scenarios.
Demerits
1. The system may require a significant amount of computing resources to perform
the clustering and classification algorithms, which could limit its scalability and
practicality for some applications.
2. Zhang [7] et al. (2022) developed a machine learning-based energy management
strategy for EVs with uncertain range anxiety. They used a combination of reinforcement
learning and deep learning algorithms to optimize the Artificial Neural Network-based
Electric Vehicle Energy Management and achieved better performance compared to
traditional rule-based strategies. However, their approach required a large amount of data
and computing resources.
Merits
1. The use of a reinforcement learning algorithm in the system can adapt to changing
driving conditions and user behavior, allowing for more efficient energy
management.
Demerits
1. The system may not be able to perform real-time energy management, which
could limit its practicality for some applications.
3. Liu et al. (2022) proposed a machine learning-based energy management strategy for
EVs with battery degradation. They used a model that could predict the degradation of the
battery and adjusted the energy consumption accordingly. Their approach improved the
battery life of the EV, but the accuracy of the battery degradation prediction was still a
limitation.
Merits
1. The use of a predictive algorithm in the system can accurately estimate the battery
degradation, allowing for more precise energy management.
Demerits
1. The system relies heavily on the accuracy of the predictive algorithm used to
estimate battery degradation. If the algorithm is inaccurate, the energy
management strategy may not be effective.
4. Shi et al. (2022) developed a machine learning-based energy management system for
EVs considering driving style and traffic conditions. They used a combination of clustering
and regression algorithms to predict the energy demand and achieved better performance
compared to traditional methods. However, their approach required a large amount of data
and computing resources.
Merits
1. The use of clustering and regression algorithms in the system can provide a more
accurate prediction of energy demand, allowing for better energy management.
Demerits
1. The system may not be able to perform real-time optimization, which could limit
its practicality for some applications.
5. Wang et al. (2021) proposed a machine learning-based energy optimization system
for EVs that could optimize the energy consumption based on the user's preferences. They
used a multi-objective optimization algorithm to balance the energy consumption and user
comfort. However, their approach required the user to input their preferences and was not
suitable for all end-users.
Merits
1. The multi-objective optimization algorithm used in the system can balance energy
consumption and user comfort, which can result in better energy efficiency and
reduced range anxiety for the driver.
3
E3S Web of Conferences 387, 04008 (2023) https://doi.org/10.1051/e3sconf/202338704008
ICSERET-2023
Demerits
1. The system may require a significant amount of computational resources to
perform the multi-objective optimization, which could limit the scalability of the
system.
3. Proposed Methodology
SmartEV is a machine learning-based energy optimization system for electric vehicles
(EVs). It is designed to optimize the charging process of EVs by utilizing real-time data on
electricity prices, weather conditions, and charging station availability. With SmartEV, EV
owners can minimize their charging costs, reduce their carbon footprint, and ensure their
EV is fully charged when they need it.
The trained model is deployed to SmartEV, an energy optimization system that can be
used in real-time for electric vehicle charging optimization. With SmartEV, EV owners can
connect their vehicle to the system and receive recommendations on the best time and
location to charge their EV based on the current electricity prices, weather conditions, and
charging station availability. By using SmartEV, EV owners can reduce their energy costs
and contribute to a more sustainable future.
𝑶𝑶𝒑𝒑𝒕𝒕𝒊𝒊𝒎𝒎𝒊𝒊𝒛𝒛𝒆𝒆𝒅𝒅 𝑬𝑬𝒏𝒏𝒆𝒆𝒓𝒓𝒈𝒈𝒚𝒚 𝑪𝑪𝒐𝒐𝒏𝒏𝒔𝒔𝒖𝒖𝒎𝒎𝒑𝒑𝒕𝒕𝒊𝒊𝒐𝒐𝒏𝒏 = 𝑴𝑴𝑳𝑳(𝑫𝑫, 𝑻𝑻, 𝑾𝑾, 𝑩𝑩, 𝑪𝑪)
Where ML represents the machine learning model that takes into account the following
variables:
D: Driver behavior, such as speed, acceleration, and braking patterns
T: Traffic conditions, including congestion and road grade
W: Weather conditions, such as temperature, precipitation, and wind
B: Battery state, including state of charge (SOC), state of health (SOH), and thermal
management
C: Charging infrastructure, such as availability and cost of charging stations
The machine learning model would use these variables to predict the optimal energy
consumption strategy for the electric vehicle. The output of the formula would be a set of
recommendations or actions that the system can take to optimize the energy consumption of
the vehicle, such as adjusting the speed, route, or charging schedule.
Data collection
Collect data related to the electric vehicle, such as battery capacity, weight, dimensions,
motor power, etc. Gather data on the vehicle usage, such as driving habits, routes, traffic
conditions, and charging history. Collect weather data, such as temperature, humidity, wind
speed, and precipitation.
Feature engineering
Extract relevant features from the collected data, such as battery state of charge, driving
speed, acceleration, deceleration, route distance, road incline, and weather conditions.
Normalize and scale the features to ensure that they are comparable and have the same
weight.
Model training
Choose a suitable machine learning algorithm, such as regression, classification, or
clustering, depending on the problem at hand. Train the model on the collected data and
feature set. Perform hyper parameter tuning to optimize the model performance.
Model validation
Validate the trained model by testing it on a different dataset, such as a cross-validation
or hold-out set. Evaluate the model's accuracy, precision, recall, F1-score, and other
performance metrics.
Optimization algorithm
4
E3S Web of Conferences 387, 04008 (2023) https://doi.org/10.1051/e3sconf/202338704008
ICSERET-2023
Develop an optimization algorithm that uses the trained model to optimize the energy
consumption of the electric vehicle. The optimization algorithm should take into account
various factors, such as battery state of charge, driving speed, route distance, road incline,
and weather conditions. Use the optimization algorithm to compute the optimal energy
consumption plan for the electric vehicle.
Implementation and deployment
Implement the optimization algorithm on a suitable platform, such as a mobile
application or an embedded system in the electric vehicle. Deploy the system and perform
real-world testing to ensure its accuracy and reliability.
Algorithm of SmartEV
Step 1: Data Collection
X = [battery SoC, distance, traffic conditions]
y = [energy required]
Step 2: Data Preprocessing
X_scaled = scale(X) # normalize data to the range [0, 1]
Step 3: Feature Engineering (none needed in this case)
Step 4: Model Selection
from sklearn.linear_model import Linear Regression
model = Linear Regression()
Step 5: Model Training
model.fit(X_scaled, y)
Step 6: Model Evaluation
y_pred = model.predict(X_scaled)
mse = mean_squared_error(y, y_pred)
Step 7: Model Optimization (none needed in this case)
Step 8: Integration and Deployment (none shown here)
4. Experimental Result
1. Accuracy
Accuracy is the degree of closeness between a measurement and its true value. The
formula for accuracy is:
(𝒕𝒕𝒓𝒓𝒖𝒖𝒆𝒆 𝒗𝒗𝒂𝒂𝒍𝒍𝒖𝒖𝒆𝒆 − 𝒎𝒎𝒆𝒆𝒂𝒂𝒔𝒔𝒖𝒖𝒓𝒓𝒆𝒆𝒅𝒅 𝒗𝒗𝒂𝒂𝒍𝒍𝒖𝒖𝒆𝒆)
𝑨𝑨𝒄𝒄𝒄𝒄𝒖𝒖𝒓𝒓𝒂𝒂𝒄𝒄𝒚𝒚 = ∗ 𝟏𝟏𝟎𝟎𝟎𝟎
𝒕𝒕𝒓𝒓𝒖𝒖𝒆𝒆 𝒗𝒗𝒂𝒂𝒍𝒍𝒖𝒖𝒆𝒆
Dataset SVM ANN SEV
100 84.12 81.37 97.67
200 80.69 85.82 94.26
300 76.62 83.54 98.21
400 74.55 79.63 95.58
500 75.94 76.72 89.87
Table 1.Comparison tale of Accuracy
The Comparison table 1 of Accuracy demonstrates the different values of existing
SVM, ANN and proposed SEV. While comparing the Existing algorithm and proposed
SEV, provides the better results. The existing algorithm values start from 75.94 to 84.12,
76.72 to 81.37 and SEV values starts from 89.87 to 97.67. The proposed method provides
the great results.
5
E3S Web of Conferences 387, 04008 (2023) https://doi.org/10.1051/e3sconf/202338704008
ICSERET-2023
Figure 3.Comparison chart of Accuracy
The Figure 3 Shows the comparison chart of Accuracy demonstrates the existing SVM,
ANN and proposed SEV. X axis denote the Dataset and y axis denotes the Accuracy ratio.
The proposed SEV values are better than the existing algorithm. The existing algorithm
values start from 75.94 to 84.12, 76.72 to 81.37 and SEV values starts from 89.87 to 97.67.
The proposed method provides the great results.
2. Precision
Precision is a measure of how well a model can predict a value based on a given input.
The precision of a model is the ratio of true positive predictions to all positive predictions.
Dataset SVM ANN SEV
100 66 75 88
200 72 73 93
300 77 68 80
400 83 71 97
500 85 62 95
𝒕𝒕𝒓𝒓𝒖𝒖𝒆𝒆 𝒑𝒑𝒐𝒐𝒔𝒔𝒊𝒊𝒕𝒕𝒊𝒊𝒗𝒗𝒆𝒆
𝑷𝑷𝒓𝒓𝒆𝒆𝒄𝒄𝒊𝒊𝒔𝒔𝒊𝒊𝒐𝒐𝒏𝒏 =
(𝒕𝒕𝒓𝒓𝒖𝒖𝒆𝒆 𝒑𝒑𝒐𝒐𝒔𝒔𝒊𝒊𝒕𝒕𝒊𝒊𝒗𝒗𝒆𝒆 + 𝒇𝒇𝒂𝒂𝒍𝒍𝒔𝒔𝒆𝒆 𝒑𝒑𝒐𝒐𝒔𝒔𝒊𝒊𝒕𝒕𝒊𝒊𝒗𝒗𝒆𝒆)
Table 2. Comparison table of Precision
The Comparison table 2 of Precision demonstrates the different values of existing SVM,
ANN and proposed SEV. While comparing the Existing algorithm and proposed SEV,
provides the better results. The existing algorithm values start from 66 to 85, 62 to 75 and
SEV values starts from 88 to 95. The proposed method provides the great results.
6
E3S Web of Conferences 387, 04008 (2023) https://doi.org/10.1051/e3sconf/202338704008
ICSERET-2023
Figure 3. Comparison chart of Precision
The Figure 3 Shows the comparison chart of Precision demonstrates the existing SVM,
ANN and proposed SEV. X axis denote the Dataset and y axis denotes the Precision ratio.
The proposed SEV values are better than the existing algorithm. The existing algorithm
values start from 66 to 85, 62 to 75 and SEV values starts from 88 to 95. The proposed
method provides the great results.
5. Conclusion
In this paper, the proposed method of SmartEV for developing a machine learning-
based energy optimization system for electric vehicles is a promising approach for
improving the efficiency and performance of electric vehicles. The system leverages
various machine learning techniques to analyze vehicle and environmental data in real-time
and make dynamic adjustments to optimize energy consumption and maximize range. The
SmartEV system is capable of adapting to different driving styles, traffic conditions, and
charging patterns to deliver personalized and efficient energy management strategies.
Additionally, the proposed approach enables the integration of renewable energy sources
and charging infrastructure to further improve the sustainability of electric mobility.
References
1. B. Liang et al., Energies, vol. 12, no. 14, p. 2714, (2019).
2. Huang, J., Tang, C., & Xie, L. (2017).. App Energy, 204, 811-820. doi:
10.1016/j.apenergy.2017.07.013
3. Wang, W., Huang, G., Guo, F., Sun, Y., & Chen, H. (2019). Energies, 12(18), 3444.
4. M. Ali, S. Ali and M. Irfan, 2021 3rd International Conf on Electrical, Communication
and Computer Eng (ICECCE), Dhaka, Bangladesh, (2021), pp. 1-5, doi:
10.1109/ICECCE51626.2021.9486464.
5. M. T. Hossain, S. S. Hasan, S. A. Khan and M. Hasan, 2021 IEEE International
Confon Artificial Intelligence and Machine Learning for Emerging Networks
(AIMLNet), Dhaka, Bangladesh, (2021), pp. 1-6, doi:
10.1109/AIMLNet51471.2021.9527671.
7
E3S Web of Conferences 387, 04008 (2023) https://doi.org/10.1051/e3sconf/202338704008
ICSERET-2023
6. Li, Y., Hu, Y., Zhu, H., Wang, H., & Zhang, X. (2022). IEEE Transactions on
Industrial Informatics, 18(1), 196-204.
7. Zhang, X., Liu, Z., Zhang, H., & Wang, C. (2022). J of Cleaner Production, 337,
130987. https://doi.org/10.1016/j.jclepro.2021.130987
8. Liu, J., Li, C., Li, Y., & Zhou, Y. (2022). J of Power Sources, 518, 230-238.
https://doi.org/10.1016/j.jpowsour.2019.02.031.
9. Shi, J., Huang, Y., Guo, X., & Zhang, Z. (2022). App Energy, 308, 117648.
10. Wang, Z., Tang, Y., Wang, L., Liu, Y., & Lu, X. (2021).. App Energy, 290, 116756.
doi:10.1016/j.apenergy.2021.116756
11. Liu, X., Zhou, Y., Cheng, X., & Huang, B. (2022). App Energy, 313, 117999. doi:
10.1016/j.apenergy.2022.117999
12. Liu, Y., Chen, H., Wang, Y., & Zhao, X. (2022). J of Power Sources, 525, 149412. doi:
10.1016/j.jpowsour.2022.149412
13. Yang, J., Zhu, Y., Xu, Z., & Chen, J. (2022). IEEE Transactions on Industrial
Informatics, 18(7), 4531-4542. doi: 10.1109/TII.2021.3105366
14. Chen, J., Huang, Z., Xue, Y., & Liu, Z. (2022). J of Power Sources, 528, 230454. doi:
10.1016/j.jpowsour.2022.230454
15. Zhang, Y., Jiang, J., Wang, Z., & Liu, Y. (2022). Jof Energy Storage, 49, 103514. doi:
10.1016/j.est.2022.103514
