Computationally Efficient Construction of Safe Configuration Space for Reconfigurable Battery Systems

Artikel i vetenskaplig tidskrift, 2026

Reconfigurable battery systems (RBSs) can enhance the operation safety and energy efficiency of, e.g., electric vehicles and grid-scale energy storage plants, by dynamically adjusting the interconnection topology of battery cells, i.e., the system configuration. However, their practical deployment faces fundamental challenges, such as the exponential growth of feasible configurations as system scale increases. Common configuration-selection approaches, predominantly relying on heuristic search, rule-based design, or partial graph analysis, cannot guarantee complete coverage of the configuration space, nor can they reliably eliminate configurations that induce unacceptable short-circuit faults or circulating currents. To address these challenges, a computationally efficient framework is proposed for constructing the safe configuration space of RBSs, enabling the identification of all electrically feasible configurations under rigorous safety constraints. Based on the sneak circuit theory, the battery system is represented by a connection matrix, and its minor expansion is employed to explicitly enumerate all safe connection paths between system terminals. Furthermore, a universal method for constructing the safe configuration space for RBSs is developed. In particular, to significantly reduce the computational complexity for path generation, a prime-encoding strategy and result reuse-based incremental computation are developed, transforming symbolic path enumeration into massively parallel GPU-based integer arithmetic while avoiding any repeated calculations. For example, in terms of path-computation efficiency for a five-cell battery system, the pruning strategy embedded in the incremental computation reduces the number of permutations to be evaluated from 10! = 3,628,800 to 182, a reduction of 99.995%. Meanwhile, the simulation-based configuration safety checking achieves an 18601× speedup over symbolic short-circuit enumeration for the five-cell battery system. The effectiveness of the proposed framework is then further validated through multiple numerical simulation studies and circuit-level experiments conducted on an RBS prototype. The results confirm complete coverage of the safe configuration space and support full-range system voltage regulation. Beyond establishing the basis for safe operation, this work also enables high-efficiency configuration search for the RBS optimal control.