RLS-MAMBA: Enhanced SOC Estimation and Parameter Identification for Lithium-Ion Batteries

Jiahui Li; Kaixin Cui; Dong Yang; Dawei Shi

doi:10.1016/j.ifacol.2025.08.070

RLS-MAMBA: Enhanced SOC Estimation and Parameter Identification for Lithium-Ion Batteries

Jiahui Li
, Kaixin Cui
, Dong Yang
, Dawei Shi^*

^*Corresponding author for this work

School of Automation

Research output: Contribution to journal › Conference article › peer-review

Abstract

This work presents the RLS-MAMBA method for lithium-ion battery State of Charge (SOC) estimation. The method integrates Recursive Least Squares (RLS) for dynamic parameter identification with the Mamba neural network for real-time error correction. RLS efficiently updates battery model parameters in real-time, ensuring accurate and adaptive parameter estimation under dynamic operating conditions. The Mamba neural network captures the nonlinear dynamics and long-range dependencies inherent in battery behavior, providing an accurate SOC estimation framework. Experiments with various initial SOC values demonstrate that this method achieves SOC estimation accuracy exceeding 97.56%, with a maximum accuracy of up to 99.98%.

Original language	English
Pages (from-to)	78-83
Number of pages	6
Journal	IFAC-PapersOnLine
Volume	59
Issue number	8
DOIs	https://doi.org/10.1016/j.ifacol.2025.08.070
Publication status	Published - 1 Jun 2025
Event	5th Joint IFAC Workshop on Control of Systems Governed by Partial Differential, Equations, CPDE 2025 and Control of Distributed Parameter Systems, CDPS 2025 - Beijing, China Duration: 18 Jun 2025 → 20 Jun 2025

Keywords

Lithium-ion battery
MAMBA
Parameter Identification
Recursive Least Squares
State of Charge Estimation

Access to Document

10.1016/j.ifacol.2025.08.070

Cite this

@article{30b1068a974d458cace6728ade6a94a6,

title = "RLS-MAMBA: Enhanced SOC Estimation and Parameter Identification for Lithium-Ion Batteries",

abstract = "This work presents the RLS-MAMBA method for lithium-ion battery State of Charge (SOC) estimation. The method integrates Recursive Least Squares (RLS) for dynamic parameter identification with the Mamba neural network for real-time error correction. RLS efficiently updates battery model parameters in real-time, ensuring accurate and adaptive parameter estimation under dynamic operating conditions. The Mamba neural network captures the nonlinear dynamics and long-range dependencies inherent in battery behavior, providing an accurate SOC estimation framework. Experiments with various initial SOC values demonstrate that this method achieves SOC estimation accuracy exceeding 97.56\%, with a maximum accuracy of up to 99.98\%.",

keywords = "Lithium-ion battery, MAMBA, Parameter Identification, Recursive Least Squares, State of Charge Estimation",

author = "Jiahui Li and Kaixin Cui and Dong Yang and Dawei Shi",

note = "Publisher Copyright: {\textcopyright} 2025 The Authors.; 5th Joint IFAC Workshop on Control of Systems Governed by Partial Differential, Equations, CPDE 2025 and Control of Distributed Parameter Systems, CDPS 2025 ; Conference date: 18-06-2025 Through 20-06-2025",

year = "2025",

month = jun,

day = "1",

doi = "10.1016/j.ifacol.2025.08.070",

language = "English",

volume = "59",

pages = "78--83",

journal = "IFAC-PapersOnLine",

issn = "2405-8963",

publisher = "Elsevier B.V.",

number = "8",

}

TY - JOUR

T1 - RLS-MAMBA

T2 - 5th Joint IFAC Workshop on Control of Systems Governed by Partial Differential, Equations, CPDE 2025 and Control of Distributed Parameter Systems, CDPS 2025

AU - Li, Jiahui

AU - Cui, Kaixin

AU - Yang, Dong

AU - Shi, Dawei

PY - 2025/6/1

Y1 - 2025/6/1

N2 - This work presents the RLS-MAMBA method for lithium-ion battery State of Charge (SOC) estimation. The method integrates Recursive Least Squares (RLS) for dynamic parameter identification with the Mamba neural network for real-time error correction. RLS efficiently updates battery model parameters in real-time, ensuring accurate and adaptive parameter estimation under dynamic operating conditions. The Mamba neural network captures the nonlinear dynamics and long-range dependencies inherent in battery behavior, providing an accurate SOC estimation framework. Experiments with various initial SOC values demonstrate that this method achieves SOC estimation accuracy exceeding 97.56%, with a maximum accuracy of up to 99.98%.

AB - This work presents the RLS-MAMBA method for lithium-ion battery State of Charge (SOC) estimation. The method integrates Recursive Least Squares (RLS) for dynamic parameter identification with the Mamba neural network for real-time error correction. RLS efficiently updates battery model parameters in real-time, ensuring accurate and adaptive parameter estimation under dynamic operating conditions. The Mamba neural network captures the nonlinear dynamics and long-range dependencies inherent in battery behavior, providing an accurate SOC estimation framework. Experiments with various initial SOC values demonstrate that this method achieves SOC estimation accuracy exceeding 97.56%, with a maximum accuracy of up to 99.98%.

KW - Lithium-ion battery

KW - MAMBA

KW - Parameter Identification

KW - Recursive Least Squares

KW - State of Charge Estimation

UR - https://www.scopus.com/pages/publications/105022694068

U2 - 10.1016/j.ifacol.2025.08.070

DO - 10.1016/j.ifacol.2025.08.070

M3 - Conference article

AN - SCOPUS:105022694068

SN - 2405-8963

VL - 59

SP - 78

EP - 83

JO - IFAC-PapersOnLine

JF - IFAC-PapersOnLine

IS - 8

Y2 - 18 June 2025 through 20 June 2025

ER -

RLS-MAMBA: Enhanced SOC Estimation and Parameter Identification for Lithium-Ion Batteries

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this