Enhancing Battery Equivalent Circuit Model Performance through Hybrid-Domain-Informed and Current-Adaptive Parameter Identification

Accurate parameter identification for battery models is essential yet challenging for intelligent battery management systems. This article proposes a new hybrid-domain-informed and current-adaptive parameter identification method for fractional-order equivalent circuit models to address the limitations of traditional time- and frequency-domain approaches. The method first utilizes frequency-domain impedance data across various state-of-charge (SOC) levels to extract polarization-related parameters, ensuring strong physical interpretability. Time-domain signals are then employed to identify ohmic and diffusion-related parameters, mitigating the challenges of time consuming of low-frequency impedance measurements and high-frequency stray parameters interference. Furthermore, time-domain data with varying current amplitudes are used to adaptively refine parameter identification, enhancing accuracy under different load conditions. The identified parameters are integrated into a battery model capable of adapting to different SOC levels and current amplitudes. The method is validated using a Li(NiCoMn)O2battery over an SOC range of 20%–80% and two aging states under diverse operating conditions. Results show that the proposed approach reduces runtime from 252 to 68 s and improves accuracy even in the presence of impedance measurement noise, reducing the root-mean-square percentage error from 5.4% to 1.9% across three distinct test conditions with varying current profiles.