IEEE Transactions on Transportation Electrification (Early Access), DOI: https://doi.org/10.1109/TTE.2026.3700443

State of Power Prediction for Series-Connected Battery Packs Considering the Inconsistency in State of Health of Cells

Simin Peng¹, Jinkang Chen¹, Daohan Zhang², Yuanliang Wang¹, Jiarong Kan¹, Aihua Tang³, Yan Ma², and Michael Pecht<sup>4

1</sup>School of Electrical Engineering, Yancheng Institute of Technology, Yancheng, China
²Department of Control Science and Engineering, Jilin University, Changchun, China
³Ministry of Education, Key Laboratory of Advanced Manufacturing Technology for Automobile Parts, Chongqing University of Technology, Chongqing, China
⁴Center for Advanced Life Cycle Engineering (CALCE), University of Maryland, College Park, MD, USA

For more information about this article and related research, please contact Prof. Michael Pecht.

Abstract:

The state of health (SOH) across a battery pack varies due to differing degradation rates among individual cells, introducing dynamic errors into state of power (SOP) predictions. A significant challenge for existing SOP prediction methods lies in simultaneously achieving cell aging identification, accurate SOP estimation, and compensation for errors arising from SOH inconsistency. To overcome these limitations, a multi-stage collaborative optimization approach for enhanced SOP prediction for series-connected battery pack is developed. The aging degrees of battery cell are quantitatively classified using fuzzy C-means clustering combined with fuzzy logic system. An adaptive step-size unscented Kalman filter is presented to improve state of charge estimation accuracy for aging cells. Accurate compensation for dynamic SOP errors is achieved by a Transformer-bidirectional long short-term memory (BiLSTM) model coupled with a Chebyshev-black Kite algorithm. Experimental results under UDDS conditions demonstrate that the proposed method achieves high SOP prediction accuracy for series-connected battery pack with a root mean square error of 0.138, which is 70.8% lower than that of grey wolf optimizer-Transformer-BiLSTM (0.472), while maintaining robust performance across different SOH levels.

This article is available online here and to CALCE Consortium Members for personal review.