A combined radial basis function and adaptive sequential sampling method for structural reliability analysis

•An adaptive sequential sampling method combined with radial basis function is proposed.•A novel active learning function can strength the fitting precision near the limit state surface.•A convergence criterion is given to terminate the sequential sampling process.•The proposed method can also be applied to other surrogate models in principle. In this paper, according to the Kriging based reliability analysis method, an efficient sequential sampling method combined with radial basis function is proposed to reduce the modeling complexity of the surrogate model and eliminate the uncertainties of the Kriging itself on the reliability analysis results. A novel active learning function is developed that can search for the sequential samples effectively among the candidate set. For terminating the sequential sampling process, a corresponding convergence criterion according to the failure probability obtained from the cross-validation method is constructed. Furthermore, the proposed method can be applied to any other surrogate model in principle. Five numerical examples demonstrate that the proposed method has high precision and efficiency as well as strong applicability in structural reliability analysis.