基于EF-U混合学习函数的自适应克里金蒙特卡罗方法OA

In structural reliability analysis,surrogate model methods aim to reduce the number of function or nu-merical model calls and lower computational cost.Effective learning functions play a crucial role in the active learn-ing process.Based on failure probability error and the sign accuracy of function values,a hybrid EF-U learning function method was proposed.During the model updating phase,the EF learning function was initially used to en-sure the overall prediction accuracy of the failure probability.When a specific transformation criterion was satis-fied,the learning function was switched to the U learning function for local refinement to improve convergence speed.Furthermore,by combining the hybrid learning function with the good lattice point sampling method and error-based stopping criteria,an improved adaptive Kriging-Monte Carlo simulation method was constructed for structural reliability analysis.Finally,the performance of the proposed method was validated through four ex-amples.The results indicate that compared with other methods,the proposed algorithm reduces the number of ac-tual function calls by 2.85%to 16.60%,decreases the failure probability error by up to 30.17%,and demon-strates better robustness.For problems with multiple failure domains,nonlinearity,and moderate dimensions,the method delivers improved computational efficiency without compromising accuracy.