Battery pack SOE update strategy for cloud-edge collaborative applications based on inconsistency assessment

Inconsistent cell performance and limited hardware computing capabilities pose significant challenges to accurate battery pack State of Energy (SOE) estimation. This paper proposes a new battery pack SOE updating strategy based on inconsistency assessment, taking the power integral as a premise and utilizing partially available charging data to achieve periodic updating of the initial SOE value. A dual-layer model framework is constructed to simultaneously realize inconsistency assessment and SOE estimation. Simulation datasets covering different inconsistency state are extensively generated based on the battery pack equivalent circuit model to achieve data enhancement. The proposed method is validated using experimental data from two battery packs. The results show accurate estimates of both the cell inconsistency distribution and SOE. The absolute error in battery pack SOE estimation across the entire charging process is less than 1 %. The proposed strategy does not require all cell voltages, and the introduction of inconsistency assessment reduces the computational cost and parameter amount of the SOE estimation model, which is able to be embedded in the on-board battery management system and has been successfully validated on the hardware system.