Spectral Representation for a Class of Non-Gaussian Processes |
| |
Authors: | Mircea Grigoriu |
| |
Affiliation: | Professor, School of Civil and Environmental Engineering, Cornell Univ., 369 Hollister Hall, Ithaca, NY?14853-3501.
|
| |
Abstract: | A new model is developed for non-Gaussian processes. The model is based on the spectral representation theorem for weakly stationary processes, can match the second moment properties and several higher order moments of any non-Gaussian process, and consists of a superposition of harmonics with uncorrelated but dependent random amplitudes. The calibration of the model to a target non-Gaussian process may require iterations. The It? formula is used to calculate higher-order moments of the proposed model. These moments can be used to tune the model such that it matches some of the higher-order moments of a target non-Gaussian process in addition to its second-moment properties. The proposed model is useful for both Monte Carlo simulation and analytical studies on the response of linear and nonlinear systems to non-Gaussian noise. Examples are presented to illustrate the use of the proposed model. |
| |
Keywords: | Gaussian process Monte Carlo method Models Stationary processes |
|
|