Bayesian-Based Parameter Identification of Single-Particle Model of Li-Ion Batteries
ECS Meeting Abstracts · 07.07.2026 · MA2026-01 (6) · 702-702
DOI: 10.1149/ma2026-016702mtgabs ↗Kurz gefasst
Fachartikel in der Zeitschrift ECS Meeting Abstracts, erschienen am 07.07.2026. Verfasst von Soomin Choi und 2 weiteren Autor:innen.
Abstract (Originalsprache)
Accurate battery modeling is essential for reliable state estimation in battery management systems (BMS) for electric vehicles (EVs). While Equivalent Circuit Models (ECM) are widely used, they often fail to provide accurate results under extreme operating conditions and require labor-intensive Hybrid Pulse Power Characterization (HPPC) tests for parameterization. Since RC parameters depend on temperature, SOC, and current, HPPC testing across a wide range of conditions is essential, rendering comprehensive parameterization increasingly challenging. Physics-based models, such as the Pseudo Two-Dimensional (P2D) model, provide high fidelity, but their computational cost and instability limit real-time applications. Single Particle Models (SPM) offer a computationally efficient alternative; however, identifying their electrochemical parameters from field data remains a significant challenge. To address these limitations, we first developed a non-iterative algorithm for SPM calculations to achieve the computational speed necessary for successful parameter identification. The SPM was also reparameterized to reduce the dimensionality of the identification process. Building on this algorithm, an iterative method for parameter identification was developed using a feedback loop that alternates between thermodynamics-based OCV estimation using the Multi-Species Multi-Reaction (MSMR) model and Bayesian optimization of the SPM parameters. This approach ensures that thermodynamic and kinetic properties are co-refined in each cycle, continuing until convergence to produce both deep-tuned OCV curves and accurate SPM parameters. The framework was validated against high-fidelity P2D simulation data under standard driving cycles (UDDS, US06). The results demonstrate that the RMSE between predicted and P2D-simulated terminal voltages decreases with each iteration, indicating successful convergence and excellent agreement with high-fidelity voltage profiles across various driving conditions. Furthermore, the identified physics-based parameters enable the analytical derivation of ECM RC parameters, effectively eliminating the need for time-consuming HPPC testing across diverse SOC, current and temperature ranges. This analytical derivation significantly accelerates the development cycle of BMS for new cell chemistries by simplifying the parameterization process. Keywords: Single Particle Model, Parameter Identification, Electric Vehicle, Open Circuit Voltage, Battery Management System
Zitieren
APA
Choi, S., Kim, M., & Kim, J. (2026). Bayesian-Based Parameter Identification of Single-Particle Model of Li-Ion Batteries. ECS Meeting Abstracts, MA2026-01(6), 702-702. https://doi.org/10.1149/ma2026-016702mtgabs
Vancouver
Choi S, Kim M, Kim J. Bayesian-Based Parameter Identification of Single-Particle Model of Li-Ion Batteries. ECS Meeting Abstracts. 2026;MA2026-01(6):702-702. doi:10.1149/ma2026-016702mtgabs
BibTeX
@article{Choi2026,
author = {Choi, Soomin and Kim, Minseo and Kim, Jinyong},
title = {Bayesian-Based Parameter Identification of Single-Particle Model of Li-Ion Batteries},
journal = {ECS Meeting Abstracts},
year = {2026},
volume = {MA2026-01},
number = {6},
pages = {702--702},
doi = {10.1149/ma2026-016702mtgabs}
}
EUVIMED