Online Learning-Based Dual Balancing of SOC and Temperature in Lithium-Ion Battery Packs

Battery management systems (BMSs) are critical for optimizing the performance and extending the lifespan of lithium-ion battery packs by maintaining uniform state of charge (SOC) and temperature across cells. While SOC balancing has traditionally been the primary focus, recent studies highlight the importance of temperature balancing to prevent accelerated degradation. This article presents a brain emotional learning (BEL)-based active balancing method that jointly regulates SOC and cell temperature within a unified, adaptive control policy designed to handle operating variability and measurement uncertainty. The proposed approach is evaluated against proportional–integral, model predictive control, and deep reinforcement learning controllers through case studies on a heterogeneous battery pack model in the MATLAB/Simulink environment, and further validated using both a virtual prototyping platform and a control hardware-in-the-loop (HIL) platform. Results demonstrate that the BEL-based approach achieves faster convergence, greater robustness to measurement noise, and improved long-term balancing performance compared with all benchmark methods.