Impact localization in anisotropic composite plates based on Bayesian inference

Abstract

The conventional impact localization strategies often assume that the wave velocity is independent of the propagation angle and obtain the wave velocity through theoretical calculation. These compromises may lead to inaccuracies of impact locations. In this paper, a Bayesian probabilistic methodology for impact localization is proposed. This approach utilizes the time of flight of diagnostic Lamb waves obtained by a piezoelectric sensor network for parameter identification. Bayes' theorem is then used to build the probabilistic relationship between measured time of flight data and unknown parameters. Finally, Markov chain Monte Carlo method is presented to implement the identification of probability distributions of impact location and wave velocity. Experimental studies carried out by dropping a steel ball on a CFRP panel are conducted to validate the proposed Bayesian impact localization strategy.