Multi-scenario state of charge adaptive estimation of lithium iron phosphate batteries based on impedance timescale information

To address the state of charge (SOC) estimation challenge in lithium iron phosphate (LFP) batteries caused by the flat open-circuit voltage plateau, a multi-dimensional feature extraction method based on impedance timescale information (TI) is proposed. TI, derived from distribution of relaxation times analysis, spans 10⁻⁵ s to several hundred seconds and reflects key dynamic processes including interfacial reactions, ion migration, charge transfer, and diffusion. Three categories of TI are defined to capture macroscopic parameter evolution, frequency-specific impedance responses, and microscopic dynamics with high precision. A dual-scenario experimental framework is designed, covering both factory-level sorting and wide-temperature calibration. SOC sampling intervals of 3 % and 5 % are compared, showing that high-resolution sampling reduces estimation error by 25.0 %. By combining feature–algorithm co-optimization with ensemble learning, temperature-adaptive SOC estimation is achieved. Among the three TI categories, feature TI delivers the best performance, with average errors of 3.53 % and 4.42 % under the two scenarios. In addition, nonlinear temperature interference on impedance features is identified, highlighting the robustness and broad applicability of the approach. This study breaks the limitations of voltage-based methods and offers an SOC estimation solution for LFP batteries that balances mechanistic insight with engineering feasibility.