Least Squares Support Vector Machine-based Advance Monte Carlo Methods for Reliability Analysis of Structures

The failure probability of structures are rather small and therefore calculation of structural reliability generally has a high computational cost. In order to reduce computational costs, this articles proposes a hybrid approach based on combination of the least squares support vector regression and two advanced Monte Carlo importance sampling and Latin hypercube sampling. Two frames and one truss example are used to evaluate the performance of the proposed algorithm. Results demonstrate that proposed method provides an accurate estimation of failure probability and that the computational costs are lower than those of other methods