Evaluating failure probability for complex engineering systems is a computationally intensive task. While the Monte Carlo method is easy to implement, it converges slowly and, hence, requires numerous repeated simulations of a complex system to generate sufficient samples. To improve the efficiency, methods based on surrogate models are proposed to approximate the limit state function. In this work, we reframe the approximation of the limit state function as an operator learning problem and utilize the DeepONet framework with a hybrid approach to estimate the failure probability. The numerical results show that our proposed method outperforms the prior neural hybrid method.

翻译：评估复杂工程系统的失效概率是一项计算密集型任务。蒙特卡罗方法虽易于实现，但收敛速度缓慢，因而需对复杂系统进行大量重复模拟以生成充足样本。为提高效率，学者们提出基于代理模型的方法来近似极限状态函数。本研究将极限状态函数近似重构为算子学习问题，并利用DeepONet框架结合混合方法估计失效概率。数值结果表明，所提方法优于先前的神经混合方法。