Reliability global sensitivity analysis method based on chaos polynomial expansion

Abstract

The invention discloses a reliability global sensitivity analysis method based on chaos polynomial expansion. The method includes the following steps that 1, the key input parameter and the key output parameter are determined according to input parameters and output parameters of a model to be analyzed; 2, chaos polynomial expansion used for representing the functional relationship between the key input parameter and the key output parameter and meeting the preset conditions is constructed; 3, the reliability numerical value of chaos polynomial expansion is calculated; 4, according to the reliability numerical value, reliability chaos polynomial expansion is constructed; 5, a reliability global sensitivity analysis index is analyzed based on the Sobol' index. The reliability global sensitivity analysis method has the advantages that calculation quantity of global sensitivity analysis can be substantially reduced, analysis results are high in accuracy and meet engineering practice better, and cross influence on global reliability by the different input parameters can be embodied.

Description

technical field [0001] The invention relates to a reliability sensitivity analysis method, in particular to a reliability global sensitivity analysis method based on chaotic polynomial expansion. Background technique [0002] Sensitivity analysis studies the effect of model output on changes in input parameters. Because of its predictive and diagnostic nature, it is often used as the first condition for modeling and model analysis. Reliability sensitivity analysis takes reliability as the research object, and mainly analyzes the extent to which the change of model input variable distribution parameters causes the change of failure probability. With the help of reliability sensitivity analysis, the factors that have a greater / smaller impact on reliability can be found, so as to provide a basis for reliability construction. Provide support for model analysis, parameter identification, reliability optimization design, etc. [0003] The traditional reliability sensitivity anal...

AI Technical Summary

Problems solved by technology

It has the following limitations: (1) It is impossible to explore the influence of the entire value space of the input parameters on reliability, so that it is impossible to find the best change area of ​​the input parameters; (2) Within the range of the probability distribution of a certain parameter, each point The partial derivatives are different, even very different, so it is inappropriate to only choose the partial derivative at a certain point in the definition domain as the sensitivity criterion; (3) when calculating the relia

Images