The Dynamic Response Index: Non-intrusive extraction of proton exchange membrane fuel cell aging signatures from 1-Hz transients

Abstract

Voltage loss is widely used to monitor proton exchange membrane fuel cell (PEMFC) aging, but it provides limited information during load-varying operation by combining irreversible degradation with reversible hydration, thermal, and recovery effects. To address this non-intrusively, a voltage-transient-based method is applied to an automotive dynamic load dataset (3076 cycles, 1008 h) using standard 1-Hz operational data. Natural current steps were treated as diagnostic excitations. A region-based comparison first identified the most stable operating segment for transient extraction. These voltage transients were decomposed into three time-domain quantities: fast apparent resistance, relaxation resistance change, and characteristic relaxation time. The analysis revealed that aging fundamentally alters the resistance–relaxation structure of the current-step response. Furthermore, comparison with periodic polarization curves showed that the 1-Hz apparent resistance growth strongly tracks the increase in steady-state polarization-derived resistance. A Dynamic Response Index (DRI) was introduced to summarize this coupled change as a relative descriptor referenced to the initial response state. Within the scope of this proof-of-concept study, the proposed method relies exclusively on routinely recorded time, current, and voltage data, extracting aging-sensitive descriptors without dedicated electrochemical characterization hardware.