AI-driven state of power prediction in battery systems: A PSO-optimized deep learning approach with XAI

Sadiqa Jafari and Yung-Cheol Byun

Energy, 2025, vol. 331, issue C

Abstract: Accurate prediction of the State of Power (SoP) in Battery Management Systems (BMS) is crucial for maximizing battery efficiency, especially in electric cars and renewable energy storage fields. SoP is a crucial metric defining lithium-ion batteries’ power efficiency. Contrary to the State of Charge (SoC), the SoP assessment suggests the battery is being utilized under demanding conditions. This research introduces a new hybrid deep learning model that combines Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM), and Convolutional LSTM (ConvLstm) architectures. The predictive performance of the hybrid method is improved even further by utilizing Particle Swarm Optimization (PSO) for hyperparameter adjustment. Explainable AI (XAI) techniques, particularly SHapley Additive exPlanations (SHAP) values, are utilized to understand model predictions and determine the crucial factors influencing SoP. The predictive models underwent training and evaluation using a dataset from NASA, focusing on lithium-ion batteries. The results demonstrate that the hybrid model, incorporating PSO optimization, achieves exceptional accuracy. It yields a Mean Absolute Error (MAE) of 0.09, Mean Squared Error (MSE) of 0.03, Root Mean Squared Error (RMSE) of 0.17, and an R2 value of 98%. Comparing these measures to individual models without PSO, there has been a noticeable improvement. The model’s transparency and dependability are increased by using SHAP values, which provide a deep knowledge of the importance of features like battery voltage and current. The study’s findings demonstrate the potential for significantly improving SoP predictions in BMS with the use of complex deep learning models optimized using PSO. This improvement can increase battery operation safety, reliability, and efficiency.

Keywords: State of power; Data-driven methods; Hyperparameter optimization; Optimization algorithms; Model interpretability; Battery systems (search for similar items in EconPapers)
Date: 2025
References: Add references at CitEc
Citations:

Downloads: (external link)
http://www.sciencedirect.com/science/article/pii/S0360544225024065
Full text for ScienceDirect subscribers only

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:eee:energy:v:331:y:2025:i:c:s0360544225024065

DOI: 10.1016/j.energy.2025.136764

Access Statistics for this article

Energy is currently edited by Henrik Lund and Mark J. Kaiser

More articles in Energy from Elsevier
Bibliographic data for series maintained by Catherine Liu ().