Accurate temperature estimation is crucial for the safe and efficient operation of lithium-ion batteries. In this work, we propose a systematic feature selection method based on ANOVA to identify impedance features primarily driven by temperature, by quantifying the effect size of different variance sources and their interactions. The ANOVA approach was first validated using synthetic data from a physics-based model, and then applied to experimental measurements. Using experimental data under dynamic conditions, the highest-ranked feature achieved a temperature estimation RMSE of 1.25 °C without requiring SoC information.

Manuel Rubio Gomez, Johannes Natterer, Marco Fischer, Johannes Thielmann, Elia Zonta, Andreas Jossen: “Statistical analysis for robust battery state estimation: Demonstrating ANOVA-driven feature selection with electrochemical impedance spectroscopy”, in: Journal of Power Sources 672 (2026) 23952, authors.elsevier.com/sd/article/S0378-7753(26)00276-