АНАЛИЗ НАДЕЖНОСТИ ЭЛЕМЕНТОВ СТРОИТЕЛЬНЫХ КОНСТРУКЦИЙ С ЯДЕРНОЙ ОЦЕНКОЙ РАСПРЕДЕЛЕНИЯ ВЕРОЯТНОСТЕЙ

Abstract

The article describes an approach to structural reliability analysis, in which distribution functions of random variables are formed on the basis of nonparametric statistics, in particular, by kernel density estimation. This approach avoids hypotheses about the choice of the distribution function of a random variable and the estimation of its parameters. The proposed approach acquires particular relevance in practical tasks of probabilistic reliability assessment during the inspection of buildings and structures when it is not possible to obtain a large amount of statistical data about the structural material or soil properties. An example of the formation of a p-box is shown on the example of probability distribution functions obtained on the basis of kernel estimation. Increasing the level of reliability in conditions of data uncertainty can be achieved by increasing the cross-sections / reinforcement of structures or by increasing the quantity and quality of statistical data, which will make the interval of the non-failure probability more informative.