Knowledge-data driven sampling diagnosis algorithm for lithium batteries on electric vehicles

Area of Science:

• Electrical Engineering
• Battery Management Systems
• Fault Diagnosis

Background:

• Voltage is a critical parameter for battery management systems (BMS).
• Voltage sampling faults can lead to operational risks and system failures.
• Investigating failure modes and developing robust diagnosis algorithms is essential.

Purpose of the Study:

• To develop a knowledge-data driven sampling diagnosis algorithm for BMS.
• To propose an online intelligent diagnosis algorithm based on fuzzy entropy and outlier detection.
• To identify and validate key features for determining voltage sampling failure modes.

Main Methods:

• Construction of a knowledge-base of failure modes using equivalent simulating models.
• Simulation of six potential BMS voltage sampling failure modes.
• Application of outlier detection with fuzzy entropy for intelligent fault diagnosis.
• Validation using fault matrix methods and battery-in-loop experiments.
• Verification using real-world data from a cloud monitoring platform.

Main Results:

• Identified symmetrical voltage distribution and near-zero voltage as key failure features.
• Validated the effectiveness of the proposed diagnosis algorithm through simulations and experiments.
• Confirmed the algorithm's ability to detect sampling line cuts in real-time.
• Demonstrated the practical applicability of the algorithm using cloud monitoring data.

Conclusions:

• The developed fault diagnosis algorithm effectively identifies voltage sampling failures in BMS.
• The knowledge-data driven approach enhances the reliability and safety of battery operations.
• The proposed methodology can be extended to other systems and cloud-based monitoring platforms.