PSO-embedded adaptive Kriging surrogate model method for structural reliability analysis with small failure probability

In the present study, a novel adaptive surrogate model method is proposed for the analysis of structural reliability with small failure probability. In order to address the problems with conventional adaptive Kriging surrogate model method based on candidate sample pool, the adaptive Kriging surrogate model method which integrates Particle Swarm Optimization algorithm (PSO) is put forward. In the course of implementation, the surrogate model is gradually improved through an iterative process and the high-value samples are selected to update the surrogate model through an optimization solution carried out by using PSO. Numerical examples are used to evaluate the computational performance of the proposed method, and a further discussion is conducted around the revision to the learning function. The results show that the introduction of PSO not only increases the possibility of obtaining high-value samples, but also significantly improves the solution accuracy of the adaptive Kriging surrogate model method for structural reliability analysis. Meanwhile, the proposed method overcomes the problem caused by the conventional candidate pool-based selection method through the optimization algorithm to determine high-value samples, achieving an excellent performance in dealing with the small failure probability. In addition, the proposed method is applicable to achieve a reasonable balance between solution accuracy and efficiency through the revised learning function which takes into account local neighborhood effects.