Extreme Estimations of Non-Gaussian Wind Pressures Integrated with Hermite Polynomial Model and Bayesian Method *

Xing-Liang Ma,Fu-You Xu

2016-01-01

Abstract:The extreme value of wind pressure is very important for the design of cladding and glass curtain wall. For estimating the extreme values of non-Gaussian wind pressure processes, a novel approach integrated with the Hermite polynomial model and Bayesian method is proposed. The Hermite polynomial model is used to formulate the PDF of stationary non-Gaussian processes. Due to the randomness of aerodynamic effect and the incomplete ergodicity of collected data, the statistics of wind pressure samples always exhibit random properties, which violate the assumption of constants. In order to consider these intrinsic random properties, the parameters involved in Hermite polynomial model are estimated by Bayesian method, in which they are considered as random variables. Moreover, the empirical knowledge and the sampled information are integrated, and the final prediction of extreme wind pressure can be more reasonable. The efficacy of the newly-proposed approach is comprehensively demonstrated and verified by comparing its estimations with the directly observed data in long-duration wind tunnel tests.