A Bayesian approach for impact load identification of stiffened composite panel

Abstract

This paper proposes a statistical approach for identifying impact location and impact force history on a stiffened composite panel using Bayesian inference, in which uncertainties from modelling error and measurement noise are explicitly included. The impact load identification problem in both space domain (impact location) and time domain (impact force history) is first transformed to a parameter identification problem by representing the impact load using a set of parameters. A forward impact model characterising the dynamic responses of a stiffened composite panel subject to a known impact load is incorporated in the identification procedure. By combining the measured data and the prior information, Bayes’ theorem is used to update the probability distributions of the parameters of impact load. In particular, Markov chain Monte Carlo method is employed for sampling the posterior distributions to estimate the impact parameters. Numerical simulation studies using noisy finite element data are conducted to demonstrate the effectiveness of the proposed method.