Efficient structural reliability analysis based on adaptive Bayesian support vector regression

To reduce the computational burden for structural reliability analysis involving complex numerical models, many adaptive algorithms based on surrogate models have been developed. Among various support vector machine regression (SVR) which is derived from statistical learning theory has demonstrated superior performance to handle nonlinear problems and avoid overfitting with excellent generalization. Therefore, take advantage of desirable features SVR, an Adaptive algorithm Bayesian SVR model (ABSVR) proposed in this study. In ABSVR, a new function devised effective selection informative sample points following concept penalty method optimization. improve uniformity design experiments (DoE), distance constraint term added function. Besides, sampling region scheme employed filter out samples weak probability density further enhance efficiency algorithm. Moreover, hybrid stopping criterion error-based using bootstrap confidence estimation developed terminate active process ensure that stops at appropriate stage. The ABSVR easy implement since no embedded optimization nor iso-probabilistic transformation required. evaluated six examples featuring different complexity, results demonstrate terms accuracy efficiency.