Active balancing strategy for AUV power battery pack based on PSO-PID algorithm

Area of Science:

• Electrical Engineering
• Materials Science

Background:

• Battery management systems require accurate state estimation.
• Inconsistent battery cells degrade overall pack performance and lifespan.
• Existing equivalent circuit models have limitations in accuracy.

Purpose of the Study:

• To develop a novel battery equalization strategy.
• To propose a fusion model for enhanced battery state estimation.
• To implement and verify an active charge equalization system.

Main Methods:

• A fusion model combining 1RC, 2RC, and PNGV equivalent circuit models using a BP neural network.
• Utilizing open-source DST dynamic operating test data for model validation.
• Developing an active equalization system controlled by a PSO-PID strategy.

Main Results:

• The proposed fusion model achieved the highest estimation accuracy (max error 0.00947, RMSE 0.00217), outperforming individual models.
• The active equalization system effectively reduced inter-cell variability in battery packs with initial SOC inconsistencies.
• The system demonstrated robustness against dynamic disturbances, maintaining low variance (average 0.0016).

Conclusions:

• The novel fusion model significantly improves battery state estimation accuracy.
• The PSO-PID controlled active equalization system is simple, effective, and superior to traditional methods, especially for increased cell inconsistencies.
• This approach enhances battery pack performance and longevity through efficient charge equalization.