arXiv subjects

Sai Hung Cheung

Publications and source records attributed to Sai Hung Cheung.

4 recordsLinked to original sources

arXiv2026

First-passage reliability sensitivity analysis of linear systems subjected to non-Gaussian wind excitations by surface decomposition method

This contribution develops a surface decomposition method for first-passage dynamic reliability sensitivity analysis of linear systems exposed to non-Gaussian wind excitations. The first-passage failure probability sensitivity is formulated as a system surface integral over a highly non-smooth and high-dimensional hypersurface. This complex integral is first decomposed into a collection of component surface integrals over the truncated smooth quadratic hypersurfaces. The dominant components are then identified based on the relative magnitudes of the first-order approximations of the component failure probabilities. A nested sampling algorithm is constructed to efficiently estimate the sum of these component surface integrals, in which the number of system limit-state function evaluations equals the number of outer-level samples while remaining independent of the inner-level sample size. A key advantage of the present approach is that the function evaluation results can be reused across different design parameters. Two numerical examples are explored to demonstrate the effectiveness of the proposed method. The results indicate that the number of function evaluations required is typically below 100 to achieve a target coefficient of variation of 0.1.

stat.ME↗

arXiv2026

Surface decomposition method for sensitivity analysis of first-passage dynamic reliability of linear systems

This work presents a novel surface decomposition method for the sensitivity analysis of first-passage dynamic reliability of linear systems subjected to Gaussian random excitations. The method decomposes the sensitivity of first-passage failure probability into a sum of surface integrals over the constrained component limit-state hypersurfaces. The evaluation of these surface integrals can be accomplished, owing to the availability of closed-form linear expressions of both the component limit-state functions and their sensitivities for linear systems. An importance sampling strategy is introduced to further enhance the efficiency for estimating the sum of these surface integrals. The number of function evaluations required for the reliability sensitivity analysis is typically on the order of 10^2 to 10^3. The approach is particularly advantageous when a large number of design parameters are considered, as the results of function evaluations can be reused across different parameters. Three numerical examples are investigated to demonstrate the effectiveness of the proposed method.

stat.ME↗

arXiv2020

Design optimization of stochastic complex systems via iterative density estimation

Reliability-based design optimization (RBDO) provides a rational and sound framework for finding the optimal design while taking uncertainties into ac-count. The main issue in implementing RBDO methods, particularly stochastic simu-lation based ones, is the computational burden arising from the evaluation of reliability constraints. In this contribution, we propose an efficient method which ap-proximates the failure probability functions (FPF) to decouple reliability. Based on the augmentation concept, the approximation of FPF is equivalent to density estimation of failure design samples. Unlike traditional density estimation schemes, where the esti-mation is conducted in the entire design space, in the proposed method we iteratively partition the design space into several subspaces according to the distribution of fail-ure design samples. Numerical results of an illustrative example indicate that the pro-posed method can improve the computational performance considerably.

stat.AP↗

arXiv2020

An efficient surrogate-aided importance sampling framework for reliability analysis

Surrogates in lieu of expensive-to-evaluate performance functions can accelerate the reliability analysis greatly. This paper proposes a new two-stage framework for surrogate-aided reliability analysis named Surrogates for Importance Sampling (S4IS). In the first stage, a coarse surrogate is built to gain the information about failure regions; the second stage zooms into the important regions and improves the accuracy of the failure probability estimator by adaptively selecting support points therein. The learning functions are proposed to guide the selection of support points such that the exploration and exploitation can be dynamically balanced. As a generic framework, S4IS has the potential to incorporate different types of surrogates (Gaussian Processes, Support Vector Machines, Neural Network, etc.). The effectiveness and efficiency of S4IS is validated by five illustrative examples, which involve system reliability, highly nonlinear limit-state function, small failure probability and moderately high dimensionality. The implementation of S4IS is made available to download at https://github.com/RobinSeaside/S4IS.

stat.OT↗