Parametric identification of systems with non-Gaussian excitation using measured response spectra

The problem of estimating parameters in dynamic systems excited by stochastic processes is addressed. Attention is focused on situations where the response processes are measurable but the excitation processes are non-Gaussian, unmeasurable and known only in terms of parameterised stochastic process models. General techniques for simultaneously estimating system and excitation process parameters are developed, based on the use of both normal, second order spectra and higher order, trispectra. The method is validated through application to some simulated data, relating to an oscillator driven by two specific kinds of non-Gaussian stochastic excitation.