Breakthrough in electric vehicle technology: Advanced SOC estimation using random forest

Beijing Institute of Technology Press Co., Ltd

 

image: 

State of charge estimation for electric vehicles using random forest

view more 

Credit: GREEN ENERGY AND INTELLIGENT TRANSPORTATION

 

In a significant leap forward for the electric vehicle (EV) industry, the accurate estimation of SOC is critical for optimizing battery usage, predicting range, and ensuring the longevity of EV batteries. Traditional methods have struggled to capture the complex, nonlinear behavior of batteries under dynamic driving conditions. This study, however, introduces the RF algorithm, which excels in real-world scenarios by leveraging decision trees and ensemble learning to form robust relationships between various input parameters，such as voltage, current, ambient temperature, battery temperature and SOC values. This innovative method promises to enhance the efficiency and reliability of EVs, addressing one of the industry's most pressing challenges.

 

The RF model significantly outperforms previous methods, including the Extreme Learning Machine (ELM), demonstrating superior accuracy and robustness. Comprehensive comparative analyses reveal that the RF model achieves a lower Root Mean Squared Error (RMSE) of 5.9028% compared to 6.3127% for ELM, and a lower Mean Absolute Error (MAE) of 4.4321% versus 5.1112% for ELM across rigorous k-fold cross-validation testing. This enhanced precision underscores the potential of RF in advancing electric mobility

 

Utilizing real-world data from 70 trips of a BMW i3 EV, the study highlights the practical application of the RF model. The integration of this SOC estimation approach into the battery management system of vehicles like the BMW i3 holds the key to more efficient and dependable EV operations. The RF model's ability to handle large datasets, robustness to noise, and feature importance analysis makes it a promising solution for the EV industry

 

The research opens new avenues for further exploration, including expanding the scope of input parameters, exploring diverse input-output configurations tailored to specific driving conditions, and incorporating feature selection techniques. These endeavors promise to further enhance the accuracy and applicability of the RF model in real-world EV applications

 

This groundbreaking study not only offers a glimpse into the future of electric mobility but also sets a new standard for SOC estimation in EVs. By harnessing the power of machine learning and advanced algorithms, the RF model promises to revolutionize battery management, improve EV range prediction accuracy, and contribute to the sustainability and efficiency of electric vehicles. As the study progresses, further research and real-world applications will likely provide additional insights and refinements, leading to even more robust and adaptable SOC estimation systems

 

Authors: Mohd Herwan Sulaiman,  Zuriani Mustaffa

Article link: https://www.sciencedirect.com/science/article/pii/S277315372400029X

 

---

Journal

Green Energy and Intelligent Transportation

DOI

10.1016/j.geits.2024.100177 

Method of Research

Experimental study

Subject of Research

Not applicable

Article Title

State of charge estimation for electric vehicles using random forest