Conditional Simulation of No-Gaussian Stochastic Process Based on Neural Networks

Abstract

In many fields such as wind engineering, ocean engineering, soil engineering and so on, it is obvious that the development of effective methods to generate sample functions of non-Gaussian stochastic processes and fields is of paramount significance for many systems subjected to non-Gaussian excitations. In this paper, neural network technique is proposed for the conditional simulation of non-Gaussian stochastic processes and fields. In machine learning of neural network, interpolation is employed to train finite non-Gaussian samples. As numerical examples, the conditional simulation of non-Gaussian fluctuating wind pressures is carried out through using back propagation neural network and generalized regression neural network respectively.