Abstract
It is a fundamental task to determine the distribution of basic random variables in structural reliability evaluation. In the present paper, a three-parameter distribution, directly defined in terms of mean value, deviation and skewness, is suggested. The new distribution can be applied (1) as a candidate distribution in fitting the statistical data of basic variables and generally presenting two-parameter distributions with small skewness, (2) to realize normal transformation and generate random samplings for random variables with unknown cumulative distribution functions in order to include them into structural reliability analysis, and (3) to provide a moment reliability index for the cases where the first-three moments of the performance function can be easily obtained. Some numerical examples are presented, the simplicity, generality and flexibility of the distribution are investigated, the applicability and efficiency of the distribution are demonstrated, and the distributions of some basic random variables are discussed.
