What question did this study set out to answer?

The study aims to improve SOC estimation accuracy in battery management systems by integrating thermal factors.

April 12, 2026SHILAP Revista de lepidopterología0 citationsOpen Access

Modeling the 3s13p Battery System With Soc Estimation Models and Machine Learning Prediction for BMS Applications

RLRina LatuconsinaJakarta State Polytechnic LLLuwis Herman Laisina

Key Points

The study aims to improve SOC estimation accuracy in battery management systems by integrating thermal factors.
Conducted experimental evaluations and mathematical modeling of battery parameters.
Applied an Adaptive Kalman Filter for real-time SOC estimation.
Used a Long Short-Term Memory neural network for time-series SOC prediction.
Performed three-dimensional thermal modeling to analyze resistance changes.
Achieved SOC estimation accuracy of ±1.5% with the Adaptive Kalman Filter.
The Long Short-Term Memory neural network produced a root mean square error of 0.95% for SOC prediction.
Energy efficiency ranged from 93% to 97%, affected by variations in internal resistance.
Thermal modeling indicated significantly increased resistance at temperatures above 45 °C.

Abstract

This study aims to enhance estimation accuracy and energy management by incorporating thermal aspects into Battery Management Systems (BMS). Through experimental evaluation and mathematical modeling, key parameters such as charging time, internal resistance, State of Charge (SOC), Depth of Discharge (DOD), and effective capacity are analyzed. The SOC profile follows a logistic curve, with system energy efficiency ranging from 93% to 97%, depending on internal resistance variations. An Adaptive Kalman Filter (AKF) is applied for real-time SOC estimation, achieving an accuracy of ±1.5%, while a Long Short-Term Memory (LSTM) neural network performs time-series SOC prediction with an RMSE of 0.95%. Furthermore, three-dimensional thermal modeling reveals a significant increase in resistance beyond 45 °C, emphasizing the effect of temperature on battery dynamics. These findings highlight the importance of integrating real-time estimation and AI-based prediction algorithms into adaptive BMS architectures, contributing to advancements in intelligent energy management for electric mobility and stationary storage systems. However, this study was conducted under controlled temperature and fixed charging conditions, which may limit generalization to dynamic real-world operations; future work will address these factors.

Demander à l'IA

Bookmark

View Full Paper