DEVELOPMENT OF A REFINED P-BOX AS A RANDOM VARIABLE MODEL IN PROBLEMS OF STRUCTURAL RELIABILITY ANALYSIS

The article describes a new type of the p-box (probability box) for random variables modeling in problems of structural reliability analysis. The advantage of the presented random variable model is that it is based on confidence estimates of statistical parameters without the need to confirm the hypothesis about the type of probability distribution function. The numerical example shows that the introduction of additional information in the form of a median’s confidence interval to the p-boxes based on the Chebyshev’s inequality is leading to reducing the area between the boundary distribution functions of the random variable. This makes it possible to reduce the level of uncertainty during further arithmetical operations with p-boxes in problems of structural reliability analysis, which is clearly visible after the discretization of the p-boxes into the Dempster-Shafer structure.