点击次数：

论文类型： 期刊论文

发表时间： 2016-10-01

发表刊物： JOURNAL OF ENGINEERING MECHANICS

收录刊物： SCIE、Scopus

卷号： 142

期号： 10

ISSN号： 0733-9399

关键字： Wind pressure; Non-Gaussian process; Inverse Johnson transformation; Numerical simulation; Peak factor

摘要： A hybrid data and simulation-based (HDSB) approach that incorporates simulated data based on an autoregression (AR) model and an inverse Johnson transformation for estimating wind pressure extremes is introduced. With the target statistical characteristics, i.e., probability density function (PDF) and power spectra density (PSD) in hand, massive simulations of non-Gaussian wind pressure can be carried out, which provides the peak factors or any desired fractile levels for design applications. Existing methods concerning the peak factors of wind pressure are briefly reviewed, and the fundamental basis and the simulation procedure of the HDSB approach are presented. The invocation of the inverse Johnson transformation in the HDSB approach enhances its applicability to the entire Pearson diagram, i.e., for all the softening and hardening stationary non-Gaussian processes with any combination of skewness and kurtosis. The efficacy of the HDSB approach in comparison with existing methods is demonstrated and verified by comparing the numerically simulated results with the directly observed data in long-duration wind tunnel tests, and with results from previous methods. The computational demand of the HDSB approach is favorable when compared to long-duration wind tunnel tests or model driven approaches. The proposed HDSB approach can serve as a building block for the development of a data-driven framework for efficiently simulating stationary non-Gaussian processes and estimating the extremes accurately. (C) 2016 American Society of Civil Engineers.