A Learning-Based Vehicle-Cloud Collaboration Approach for Joint Estimation of State-of-Energy and State-of-Health.

Area of Science:

• Electrical Engineering
• Artificial Intelligence
• Materials Science

Background:

• Accurate estimation of battery state-of-energy (SOE) and state-of-health (SOH) is critical for electric vehicle (EV) battery management systems (BMS).
• EV complexity and environmental variability pose significant challenges to traditional estimation methods.
• On-board computational limitations restrict the use of complex machine learning algorithms for real-time battery state estimation.

Purpose of the Study:

• To propose a novel joint SOE and SOH prediction algorithm for EVs utilizing a vehicle-cloud collaboration framework.
• To develop a high-precision estimation method that overcomes the limitations of on-board computation.
• To enable accurate battery state prediction using cloud-based historical data and on-board processing.

Main Methods:

• A hybrid deep learning model combining Long Short-Term Memory (LSTM), Bi-directional LSTM (Bi-LSTM), and Convolutional Neural Networks (CNNs) was employed.
• Bayesian optimization was integrated with Bi-LSTM for robust state-of-health (SOH) prediction using performance degradation indicators.
• CNN-LSTM models were utilized for direct, nonlinear mapping of state-of-energy (SOE), avoiding complex parameter identification.

Main Results:

• The proposed vehicle-cloud approach achieved a joint estimation accuracy with errors maintained within 3% on the NASA battery dataset.
• The method successfully integrated SOH prediction with SOE estimation through SOH correction, enabling joint estimation across different time scales.
• Direct mapping models for SOE proved effective under complex operating conditions, bypassing traditional parameter updating requirements.

Conclusions:

• The vehicle-cloud collaboration strategy offers a feasible and effective solution for high-precision joint estimation of battery SOE and SOH in EVs.
• This approach leverages cloud computing for complex analysis (SOH prediction) and enables real-time SOE correction, enhancing overall battery management.
• The proposed method demonstrates significant promise for future advancements in intelligent BMS for electric vehicles.