Active Sensing Acousto-Ultrasound SHM via Stochastic Non-stationary Time Series Models

In this work, a novel statistical approach for damage detection and identification in the context of ultrasonic guided wave-based damage diagnosis is proposed using stochastic functional series time-varying autoregressive (FS-TAR) models. Wavelet functions are used as the functional basis family and the coefficients of projection of the time-varying model parameters are estimated via a maximum likelihood scheme. Damage detection and identification are tackled within a statistical decision making framework while appropriate thresholds are derived using pre-determined type I error probability levels. Both damage intersecting and non-intersecting, with respect to wave propagation, paths are considered in a multi-sensor aluminum plate in pitch-catch configuration. The method’s robustness, effectiveness, and limitations are discussed. The results indicate the effectiveness of the proposed method in detecting and identifying damage within a statistical setting.