Overcoming low-frequency non-stationarity in battery impedance spectroscopy through optimized multisine excitation

Abstract The exponential growth of electric vehicle and grid storage markets demands rapid, accurate battery diagnostics that current impedance spectroscopy cannot deliver. While sine sweep methods provide precise characterization, they suffer from time-invariance violations below 1 Hz due to extended measurement durations that destabilize battery state, creating a significant bottleneck for real-time applications. We present the first systematic optimization framework for multisine excitation based on Evolutionary Role Playing Game Theory (ERPGT) that simultaneously addresses speed and accuracy constraints. Our method achieves unprecedented performance: 4.34× faster measurements while delivering 71.8 % reduction in magnitude deviation and 83 % improvement in phase accuracy over conventional approaches. The optimized signals maintain RMS deviations below 0.005 % across 0.1–1.0 A ranges, enabling precise sub-Hz characterization critical for battery aging analysis. These advances unlock real-time battery diagnostics for automotive and grid applications, enabling predictive maintenance strategies that could extend battery lifetime while reducing diagnostic costs by an order of magnitude.