ASSESSMENT OF A METHOD FOR ESTIMATING PEAKS OF STATIONARY RANDOM PROCESSES BY USING EXTREME VALUE TYPE I DISTRIBUTIONS
Dat Duthinh, Dong Hun Yeo, Emil Simiu
The purpose of this note is to assess a method based on the Extreme Value Type I distribution currently employed in wind engineering practice to estimate stationary time series peaks. The note considers the interesting case of time series with Gaussian marginal distribution for which a closed form solution exists. The solution is based on the classical mean zero upcrossing approach and makes it possible to determine peak statistics unaffected by sampling, measurement, or probabilistic modeling errors. The assessment is performed in a transparent manner by comparing results of this method with those obtained by using the classical mean zero upcrossing rate approach. It is shown that current EV I-based method can result in the underestimation by approximately 10 % to 13 % of expected random process peaks used for structural design purposes.
Journal of Structural Engineering-ASCE
Extreme Value Type I distribution, mean zero upcrossing rate, non-Gaussian process, normal random process, random process peaks, stationary time series.
, , D.
and Simiu, E.
ASSESSMENT OF A METHOD FOR ESTIMATING PEAKS OF STATIONARY RANDOM PROCESSES BY USING EXTREME VALUE TYPE I DISTRIBUTIONS, Journal of Structural Engineering-ASCE
(Accessed May 28, 2023)