Speaker
Description
Modern engineering structures are developing toward larger scales, greater complexity, and harsher environments, posing significant challenges to structural reliability analysis. Traditional methods often suffer from high computational costs and limited accuracy when dealing with highly nonlinear problems, multidimensional uncertainties, and expensive simulations. To address these challenges, this study develops an efficient and accurate framework for structural reliability analysis by integrating optimization algorithms, active learning, multifidelity information fusion, ensemble modeling, and physics-informed learning. The main contributions are as follows: (1) A first-order reliability method based on an augmented Lagrangian penalty function and an enhanced sparrow search algorithm is proposed to improve convergence and avoid local optima and numerical ill-conditioning in complex nonlinear reliability problems. (2) A hybrid active-learning strategy is developed for Kriging modeling, balancing global exploration and local exploitation while directing samples toward high-risk and informative regions. (3) A multifidelity active Kriging method is established by effectively integrating computationally efficient low-fidelity models with accurate high-fidelity simulations, significantly reducing computational costs. (4) An entropy-weighted ensemble surrogate model is developed to improve prediction stability and accuracy by adaptively combining multiple surrogate models according to their performance and uncertainty. (5) A physics-informed neural network-based reliability and optimization framework is proposed, embedding governing equations and boundary conditions into the loss function to enforce physical constraints and ensure physically feasible optimal solutions. The proposed methods provide effective tools for reliability-based design of complex engineering systems, including deep-water risers.
| Topic Area | Fatigue, Corrosion and Structural Integrity of Engineering Structures and Mechanical Systems |
|---|