A Bayesian approach for damage localization in plate-like structures using Lamb waves

This paper presents an investigation of the applicability of a Bayesian system identification theory for localizing damage in plate-like structures, while considering the uncertainties from modeling and measurement. Diagnostic Lamb waves are excited and received by a piezoelectric sensor network before and after damage to obtain scattered waves that contain characteristic information about the damage. After the time-of-flight (ToF) of the scattered waves in each actuator-sensor path is measured by a continuous wavelet transform (CWT), a Bayesian approach is developed to identify the damage location and wave velocity. By combining the prior information and the measured ToF data, Bayes' theorem is used to update the probability distributions of the parameters about the damage location and wave velocity. In particular, a Markov chain Monte Carlo (MCMC) method is employed for sampling the posterior distributions of the unknown parameters. A numerical study for an aluminum plate and experimental studies for a stiffened aluminum panel and a composite laminate are conducted to validate the proposed Bayesian damage localization approach.