Non-sampling inverse stochastic numerical-experimental identification of random elastic material parameters in composite plates

A non-sampling probability identification method based on the generalized polynomial chaos (gPC) expansion is adopted for estimating random parameters of composite plates form experimental eigenfrequencies. For that, the parameters and the eigenfrequencies are approximated using gPC expansion. Distribution functions of the eigenfrequencies are identified from experimental data employing the Bayesian inference. This identification is then used to construct a vector of random variables and an orthogonal basis for eigenfrequency expansions. The parameters are characterized by the gPC having unknown deterministic coefficients and the same random basis as the eigenfrequencies. The stochastic finite element simulation of the plates bears as the model from which the parameter coefficients are estimated via an inverse problem. The major advantage of the method is using deterministic identification procedure. An application is presented for which samples of orthotropic laminated plates are tested to identify E-moduli, shear modulus and the major Poisson's ratio from measured modal frequencies.