Simulation of Stochastic Processes by Spectral Representation

TLDR

The paper focuses on simulating one‑dimensional stationary Gaussian stochastic processes via spectral representation. The method is intended for Monte Carlo solutions of stochastic problems in engineering mechanics and structural engineering. Sample functions are generated efficiently with a cosine series derived from spectral representation. The generated sample functions accurately reproduce the target power spectral density, autocorrelation, and ergodic properties, become asymptotically Gaussian as the series length increases, and can be computed efficiently with FFT.

Abstract

The subject of this paper is the simulation of one-dimensional, uni-variate, stationary, Gaussian stochastic processes using the spectral representation method. Following this methodology, sample functions of the stochastic process can be generated with great computational efficiency using a cosine series formula. These sample functions accurately reflect the prescribed probabilistic characteristics of the stochastic process when the number N of the terms in the cosine series is large. The ensemble-averaged power spectral density or autocorrelation function approaches the corresponding target function as the sample size increases. In addition, the generated sample functions possess ergodic characteristics in the sense that the temporally-averaged mean value and the autocorrelation function are identical with the corresponding targets, when the averaging takes place over the fundamental period of the cosine series. The most important property of the simulated stochastic process is that it is asymptotically Gaussian as N → ∞. Another attractive feature of the method is that the cosine series formula can be numerically computed efficiently using the Fast Fourier Transform technique. The main area of application of this method is the Monte Carlo solution of stochastic problems in engineering mechanics and structural engineering. Specifically, the method has been applied to problems involving random loading (random vibration theory) and random material and geometric properties (response variability due to system stochasticity).