Simulation of r Highest-Order Extreme Values of Correlated Random Process

Several existing methods on the extreme value estimation and signal simulation are reviewed in this study with evaluation of their performances in the simulation of the r highest-order extreme values. The limitations of these methods are also discussed. Then, a new method to simulate the r highest-order extreme values of a random process is presented with improvement to these limitations. Extreme value theory is involved in the estimation and Monte Carlo simulation is used in the numerical study. Estimation of the annual top 10 highest 10-min mean wind speeds is conducted for illustration. Results show that the top-order extreme values of the correlated random processes can be well simulated by the proposed method. (C) 2015 American Society of Civil Engineers.