Accelerating Frequency Response Measurements Through Sparse Data‐Driven Reconstruction Using the Loewner Framework

ABSTRACT A data‐driven approach based on the robust Loewner framework is proposed to identify the minimum amount of frequency‐domain information required for accurate system characterization. By constructing reduced‐order state‐space models directly from measured frequency response data, this method determines the effective system order and enables reliable reconstruction of the transfer function and the corresponding distribution of relaxation times from only a sparse set of informative frequency samples. The methodology is applied on concentration‐alternating frequency response analysis of polymer electrolyte membrane fuel cells, where square‐wave oxygen concentration perturbations naturally generate multi‐harmonic excitation signals. This strategy is generally applicable to linear frequency response techniques and provides a practical route toward online monitoring and accelerated diagnostics of a broad range of electrochemical systems.