A Novel Sampling Method Based on Normal Search Particle Swarm Optimization for Active Learning Reliability Analysis

In active learning reliability methods, an approximation of limit state function (LSF) with high precision is the key to accurately calculating failure probability (Pf). However, existing sampling methods cannot guarantee that candidate samples can approach LSF actively, which lowers accuracy and stability results causes excess computational effort. this paper, a novel samples-generating algorithm was proposed, by group evenly distributed points on predicted performance (either real one or surrogate model) could be obtained. proposed method, determination considered as optimization problem in absolute value objective function. After this, normal search particle swarm (NSPSO) designed deal such problems, consists pattern multi-strategy framework ensures uniform distribution diversity solution intends cover optimal region. Four explicit functions two engineering cases were employed verify effectiveness NSPSO method. state-of-the-art multi-modal algorithms used competitive methods. Analysis show method outperformed all provide LSF.