Bayesian Inference Method to Identification of Random Parameters
* Presenting author
The dynamic and static behavior of the fiber reinforced composite is uncertain in nature due to the inherited randomness in elastic moduli, damping parameter, random fiber orientation, etc. Non-sampling based stochastic methods like, generalized Polynomial Chaos (gPC) expansion is used to model this uncertainty much better way due to its computational efficiency. The gPC expansion involve to construct the orthogonal polynomials of standard random variables with deterministic coefficients. To construct the orthogonal polynomial it is needed to be identify the probability distribution of the random variable which can be empirically identified through the Pearson model or previous references. Identification of the prior distribution always not be best fitted with the experimentally observed data. So it is needed to be identified a suitable model class that can be fitted to experimental data better way. The selection of the best model class can be address by the inference based on the Bayes’ Theorem. The success of Bayesian method lies to construct a good the Posterior distribution after taking the less experimental data into account. The aim of this work is to construct an optimal distribution of the random parameters with lesser experimental data through the Bayesian inference.