A novel AR-MEM-PJTM method for simulating multivariate stationary non-Gaussian wind pressure processes

Abstract

It is generally accepted that the wind-induced response time domain analysis for nonlinear structures requires accurate and fast simulation of non-Gaussian wind pressures. Recently, an enhanced autoregressive (AR)-based method for simulating univariate wind pressures has been proposed by some authors of this study. However, the corresponding method for simulating multivariate wind pressures is missing. This study comprehensively uses AR, maximum entropy method (MEM), piecewise Johnson transformation model (PJTM) and proposes a novel AR-MEM-PJTM method for simulating multivariate non-Gaussian wind pressures. In this method, a set of closed-form formulations for estimating higher-order moments of the AR's input process vector are firstly theoretically derived. Next, MEM is used to approximate the marginal probability distribution function of the input process vector, which is then applied to determine PJTM. The proposed AR-MEM-PJTM method is illustrated in the numerical examples to be capable of considering more moments, thus result in satisfactory simulations for a variety of multivariate non-Gaussian wind pressures. It is also pointed out that the proposed method is not restricted by the application range, which actually exists in the conventional methods using AR model. Note that the proposed method can also be applied to simulate other non-Gaussian processes such as the wind speed.