High-dimensional reliability-based design optimization using stochastic emulators
The paper addresses computationally expensive reliability-based design optimization by replacing traditional nested loops with a unified stochastic representation. It constructs stochastic emulators like Generalized Lambda Models or Stochastic Polynomial Chaos Expansions to approximate complex conditional response distributions. This allows for semi-analytical evaluation of failure probabilities without needing time-consuming Monte Carlo simulations, leading to substantial computational gains in high-dimensional problems.