Circulating-Current-Impedance-Based Adaptive Assessment of Inconsistency in Energy Storage Systems

Cell inconsistency constrains the usable capacity, operational safety, and service life of large-scale battery energy storage systems. Most online methods infer inconsistency from voltage, current, or temperature; in parallel branches, however, self-balancing currents redistribute the operating current and can obscure its external signatures. This paper proposes a mode-adaptive assessment method based on circulating-current impedance. Circuit analysis reveals how open-circuit-voltage and resistance differences reverse the circulating-current direction among discharge, idle, and charge. The signed branch response is combined with each cell voltage to form a topology-conditioned cell-level quasi-impedance, which is mapped to a 0–100 consistency score by mode-specific rules. In controlled 6P3S experiments, the predefined resistance-disturbed cells received the two lowest scores in both same-cluster and cross-cluster scenarios; on the same 18-cell records, a baseline-corrected peak-temperature-rise feature did not place either disturbed pair among the two highest-ranked cells. Update times remained at the millisecond level for both 18- and 1200-cell configurations. A field application produced 17.328 million valid cell–window evaluations and persistent cluster-level screening patterns. These results support a physically interpretable and computationally lightweight approach to topology-conditioned cell inconsistency screening. Because independent fault labels were unavailable, the field evidence establishes operational scalability rather than diagnostic accuracy.