Surface Temperature Assisted State of Charge Estimation for Retired Power Batteries

Area of Science:

• Battery Technology
• Electrochemistry
• Thermal Management

Background:

• State of Charge (SOC) estimation for retired batteries is challenging due to degradation and aging.
• Traditional electrical-parameter-only methods are inaccurate for aged batteries because of altered internal resistance and heat generation.
• Heterogeneous aging mechanisms in retired batteries complicate SOC prediction.

Purpose of the Study:

• To develop an accurate SOC estimation method for retired power batteries using surface temperature data.
• To leverage infrared thermal imaging to identify key temperature features correlated with SOC.
• To integrate thermal and electrical data for improved SOC estimation in aged batteries.

Main Methods:

• Infrared thermal imaging to capture temperature features from positive/negative tabs and central areas.
• Development of a Gated Recurrent Unit (GRU) neural network to process multi-region temperature and electrical data.
• Experimental validation under constant current discharge conditions (0.5 C-2 C).

Main Results:

• The proposed surface temperature-assisted method significantly reduces Root Mean Square Error (RMSE) by 64.3-68.1% compared to electrical-parameter-only methods.
• Achieved RMSE values below 1.04 across all tested discharge rates, outperforming traditional voltage-based models.
• The GRU model effectively captures spatial-temporal thermal-electrical interactions and hysteresis effects in aged batteries.

Conclusions:

• Surface temperature-assisted SOC estimation is a more accurate approach for retired batteries.
• The GRU model integrating thermal and electrical data enhances reliability for secondary battery applications.
• This method improves the performance of retired battery management systems in energy storage applications.