Damage detection and localisation based on first passage time histograms

Abstract

Structural health monitoring has become increasingly critical as civil infrastructure continues to age. This study presents a novel application of empirical cumulative distribution functions (ECDFs) of first passage times (FPTs) for structural damage detection and localisation. FPTs were efficiently computed from vibration data over a wide range of initial and target amplitude levels, producing an FPT map that encodes rich dynamic information. Laboratory experiments were conducted on a scaled structure excited by band-limited white noise and monitored using a laser Doppler vibrometer. Controlled damage was introduced, and ECDFs of FPTs were compared between healthy and damaged states using the Kolmogorov–Smirnov, Wasserstein, and a newly proposed two-sample test. The ECDF map provided a three-dimensional representation of the system response, offering greater sensitivity to subtle damage effects than traditional indicators such as natural frequency shifts or mode shape variations. For localisation, a numerical twin model was calibrated from the experimental setup and used to simulate various damage scenarios. Comparison between experimental and simulated FPT maps enabled accurate identification of damage locations, including cases near supports where modal properties changed only marginally. The results demonstrate that ECDFs of FPTs constitute a powerful and sensitive metric for both detecting and localising structural damage.