Numerical modeling and higher moments analysis of response characteristics in random structural systems

A moment-based method founded on the maximum-entropy principle is applied in this paper with the aim of quantifying the uncertainty associated with structural behaviour, for systems affected with imperfect knowledge in the input parameters. A numerical model implemented on a microcomputer is employed for the preliminary identification of a response surface approximating the input-output mechanical model. It is shown how this initial stage enables the subsequent analysis of uncertainty propagation to be supplied with sufficient information about the probability distribution of the system output.