The Quantification of Uncertainty in SHM Features Derived From Transmissibility Estimation

This paper develops uncertainty models for various an estimator used in computing the frequency-domain transmissibility function for single-input-multiple-output (SIMO) linear systems. The uncertainty is assumed to come from both internal (estimator) and external (environmental, noise, etc.) sources, and it is propagated through the estimation process to arrive at closed-form probability density functions for the estimation magnitude. The models is validated on a bolted plate in the laboratory, and receiver operating characteristics (ROCs) are computed to evaluate performance in terms of meaningful metrics such as probability of detecting a structural change within the context of structural health monitoring.