Active learning Kriging method based on particle swarm optimization for reliability analysis with random and interval hybrid uncertainty

We propose an active learning Kriging reliability method, based on the particle swarm optimization algorithm, to solve structural reliability assessment problems in which both random variables and parameter interval uncertainty coexist. The method optimizes the selection of optimal training samples by using the U learning function as the optimization objective, combined with search space reduction and domain truncation techniques. An error-based stopping criterion is employed to ensure termination of the algorithm. The effectiveness and feasibility of the proposed method are validated through five specific examples, whose results demonstrate the capability of the method to improve accuracy and computational efficiency in the reliability assessment.