A method for the evaluation of the statistical moments of linear one-d
imensional or multi-dimensional systems subjected to non-Gaussian inpu
t in polynomial forms of filtered normal or non-normal delta-correlate
d processes is presented. The proposed procedure allows one to avoid t
he solution of a set of non-linear differential equations, requiring t
he solution of three sets of linear differential equations. The latter
sets are: the equations governing the evolution of the moments of the
response forced by input-output cross moments; the equations useful f
or the evaluation of the input-output cross-moments and the equations
governing the evolution of the statistical moments of the filtered inp
ut.