报告人简介：

Dr. Marcos Valdebenito is Senior Scientist at the Chair for Reliability Engineering at TU Dortmund University, Germany. He obtained his doctoral degree in Civil Engineering from University of Innsbruck, Austria, after completing his engineering and master’s studies at Santa Maria University, Chile. His research focuses on uncertainty quantification, structural reliability, stochastic dynamics, Bayesian updating, surrogate modelling, and reliability-based design optimization in computational mechanics. He has authored more than 100 journal publications. Dr. Valdebenito received the K.J. Bathe Award in 2016 and a research fellowship from the Alexander von Humboldt Foundation. He currently serves on the editorial boards of the international journals Computers & Structures, Structural Safety, and Machine Learning for Computational Science and Engineering.

报告简介：

First-excursion probabilities constitute a fundamental measure for quantifying the reliability of engineering systems subjected to random dynamic loading. For linear dynamical systems with deterministic structural properties and Gaussian stochastic excitation, a variety of efficient simulation methods have been developed and successfully applied. In many practical applications, however, structural parameters such as masses, stiffnesses, or damping coefficients are themselves uncertain. The resulting reliability problem involves both excitation uncertainty and parameter uncertainty, leading to a substantial increase in computational complexity.

This presentation introduces a simulation framework for the efficient estimation of first-excursion probabilities of linear dynamical systems with uncertain structural parameters. The proposed approach combines active-learning Gaussian process regression, importance sampling, and multidomain Line Sampling. A surrogate model is constructed in the space of uncertain structural parameters and is employed to identify regions that contribute most significantly to failure. Active learning is used to adaptively refine the surrogate only where required. The resulting model is subsequently used to construct an importance sampling density for the structural parameters, while multidomain Line Sampling is employed to efficiently account for stochastic excitation.

The proposed methodology preserves the accuracy of simulation-based reliability analysis while substantially reducing the associated computational effort. Numerical examples involving stochastic structural dynamics are used to demonstrate the performance of the approach and to illustrate the benefits of combining surrogate modelling, active learning, and advanced simulation techniques for rare-event estimation.