Deep learning-based state of charge estimation for electric vehicle batteries: Overcoming technological bottlenecks

Area of Science:

• Electrical Engineering
• Artificial Intelligence
• Sustainable Transportation

Background:

• Accurate State of Charge (SOC) estimation is crucial for electric vehicle (EV) battery management and efficiency.
• Traditional SOC estimation methods face challenges with complex influencing factors.
• Enhancing EV performance requires advanced battery management strategies.

Purpose of the Study:

• To develop and validate a novel deep learning approach for precise EV battery SOC estimation.
• To improve EV efficiency and battery management through advanced estimation techniques.
• To leverage real-world driving data for robust SOC prediction.

Main Methods:

• A deep learning model was trained using a diverse dataset including environmental, vehicle, and battery parameters.
• Real-world driving data from a BMW i3 electric vehicle was integrated into the model.
• The model's performance was validated through 72 tests incorporating 25 environmental variables.

Main Results:

• The proposed deep learning model demonstrated superior accuracy and reliability in SOC estimation compared to traditional methods.
• The model effectively captured intricate dynamics influencing SOC by analyzing interrelationships among various factors.
• Validation using real-world driving data confirmed the model's feasibility and effectiveness.

Conclusions:

• The study presents a significant advancement in EV battery SOC estimation techniques.
• The findings offer actionable insights for improving EV efficiency, energy conservation, and carbon reduction.
• This deep learning approach addresses a pivotal challenge in EV battery management, contributing to sustainable transportation.