CSP-free adaptive Kriging surrogate model method for reliability analysis with small failure probability (2304.07010v5)

Abstract: In the field of reliability engineering, the Active learning reliability method combining Kriging and Monte Carlo Simulation (AK-MCS) has been developed and demonstrated to be effective in reliability analysis. However, the performance of AK-MCS is sensitive to the size of Candidate Sample Pool (CSP), particularly for systems with small failure probabilities. To address the limitations of conventional AK-MCS that relies on CSP, this paper proposes a CSP-free AK-MCS. The proposed methodology consists of two stages: surrogate model construction and Monte Carlo simulation for estimating the failure probability. In the stage of surrogate model construction, the surrogate model is iteratively refined based on the representative samples selected by solving the optimization problem facilitated by Particle Swarm Optimization (PSO) algorithm. To achieve an optimal balance between solution accuracy and efficiency, the penalty intensity control and the density control for the experimental design points are introduced to modify the objective function in optimization. The performance of the proposed methodology is evaluated using numerical examples, and results indicate that by leveraging an optimization algorithm to select representative samples, the proposed CSP-free AK-MCS overcomes the limitations of conventional CSP-based AK-MCS and exhibits exceptional performance in addressing small failure probabilities.