State of Charge Estimation of EV Secondary Battery Pack Using Hybrid Hedge Feedforward Feedback‐Based Gated Recurrent Unit to Extend Lifespan

ABSTRACT Accurate estimation of state of charge (SoC) and maintaining balanced charge levels across secondary battery cells are crucial in battery management systems (BMSs) to extend battery life while improving the performance and thermal stability of Li‐ion batteries (LIBs) in electric vehicles (EVs). However, there are still underexplored challenges associated with circulating currents in electrochemical cells during continuous operation which can overheat battery packs, reducing their life span or result in dangerous thermal runaways. This paper investigates SoC estimation using various real‐world charging and discharging profiles, along with charge‐balancing strategies to enhance the longevity of parallel‐connected Li‐ion battery cells. A newly developed hedge feedforward feedback‐based gated recurrent unit with H∞ controller (HFF‐GRU‐H∞) is introduced to improve the SoC estimation accuracy with comparisons to nine widely‐applied deep‐learning algorithms. Moreover, SoC balancing for three individual battery cells is achieved using a bidirectional DC/DC power converter controlled by an H∞ robust control system during charging‐discharging cycles. The experimental results indicate that SoC capacity estimation error can be reduced to 0.043%. Also, the applied optimization algorithm minimized the determination time to 0.477 s when benchmarked with existing methods leading to better charge balance among the battery cells. As a result, the overall battery pack lifespan can be extended by 27.7%, offering substantial advantages for industrial applications.