Dynamic Cell Modeling for Accurate SOC Estimation in Autonomous Electric Vehicles

Ajao, Qasim; Sadeeq, Lanre · 2023 · Crossref

DOI: 10.4236/jpee.2023.118001

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper addresses the challenge of accurate State-of-Charge (SOC) estimation for Lithium-Ion Polymer (LiPB) battery cells in autonomous electric vehicles (AEVs). The research is motivated by the harsh operating conditions of AEVs, which involve high current rates up to ±25C, highly dynamic rate profiles, and temperatures ranging from −30°C to 50°C. These conditions significantly exceed those of portable electronics, rendering standard SOC estimation methods insufficient. The authors argue that precise SOC estimation requires a detailed cell model that treats SOC as a state variable within a discrete-time state-space framework, enabling the use of Kalman filtering for dynamic estimation. The study evaluates multiple cell modeling approaches, progressing from simple to complex structures. Initial models utilize Coulomb counting for the state equation and empirical rules (such as Shepherd’s or Nernst models) for the output equation, though these lack the ability to predict cell relaxation dynamics. To address this, the authors propose an improved model incorporating additional filter states to account for relaxation and other dynamics in the closed-circuit cell voltage. The most advanced method combines nonlinear autoregressive filtering with dynamic radial basis function networks. The methodology relies on non-invasive measurements of terminal voltage, current, and temperature, as the battery management system lacks direct control over the vehicle’s power demands. Model parameters are determined through system identification using laboratory test data from 8 Ah LiPB cells optimized for power-intensive applications. The results demonstrate that the proposed models, particularly those incorporating relaxation dynamics and nonlinear filtering, achieve high accuracy in predicting cell behavior. The most accurate models yielded a root-mean-square (RMS) estimation error lower than the quantization noise floor expected in battery management system designs. By treating SOC as a state variable, the Kalman filter provides not only a point estimate of SOC but also an uncertainty bound (e.g., 55% ± 7%). This capability allows the vehicle controller to utilize the battery pack’s full operating range with confidence, avoiding the risks of overcharging or undercharging. The significance of this work lies in its provision of a robust, generalizable modeling framework for high-power battery systems. Unlike electrochemical or impedance-based methods, which are either too complex for real-time implementation or require invasive signal injection, the proposed approach is suitable for real-time application in AEVs. The ability to accurately estimate SOC and its uncertainty under extreme dynamic conditions ensures safer and more efficient operation of autonomous electric vehicles, addressing a critical gap in existing battery management technologies.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• situational awareness