Sensor-less estimation of battery temperature through impedance-based diagnostics and application of DRT

Abstract

Temperature has a substantial influence on the overall safety and performance of Lithium-ion batteries. Given the constraints of onboard thermal sensors and their inability to accurately measure internal cell temperature, a reliable temperature estimation has become a crucial aspect of battery state monitoring. This study exploits the temperature-sensitivity of electrochemical impedance spectroscopy (EIS) measurements to propose a sensor-less method to accurately estimate the internal temperature of commercial lithium-ion batteries. The presented study explores the reliability and limitations of the EIS-based method via a comparative analysis on two different cell types, i.e., high impedance (cylindrical 5 Ah) and low impedance (pouch 40 Ah) cells, over a range of multiple SOCs and temperatures. Furthermore, a novel approach of distribution of relaxation times (DRT) to extract the temperature-sensitive features from EIS data is also investigated. The results show that method is capable of estimating the internal temperature of high-energy cylindrical cells with an accuracy of ±0.41 °C, and high power pouch cells with an accuracy of ±2.22 °C over the entire range of tested SOCs. Overall, the Arrhenius model (for both cell types) represents a good fit for all the extracted features with R² > 0.9. Charge transfer resistance (R_CT) was found to be the most significant predictor for cylindrical cells and ohmic resistance (R_ohm) for pouch cells. Furthermore, DRT peak heights can serve as a thermally sensitive feature for cell temperature estimation with good accuracy (typically <3 °C, though dependent on cell impedance response profile), and potential for broader applicability than features derived from equivalent circuit modelling. The study illustrates the opportunities and challenges associated with implementing impedance-based temperature estimation methodologies.