工程可靠性与随机力学国际联合研究中心

10 周年庆典青年讲座

(Youth Lectures for the 10th Anniversary

Celebration of JCERSM) 第 1 期

工程可靠性与随机力学国际联合研究中心

2026 年第 3 期（总第 119 期）学术报告

工程力学研究中心第 74 期学术报告

同济大学土木工程学院院级高等讲堂

报告主题

TOPIC

高维非线性随机动力系统仿真的时滞状态空间代理模型

Time-delay state space surrogate model for emulating high dimensional
nonlinear stochastic dynamical systems

报告人

SPEAKER

Dr. Kai Cheng（成凯）
Alexander von Humboldt Research Fellow at the Engineering
Risk Analysis Group, Technical University of Munich (TUM)

报告时间

TIME

2026 年 4 月 13 日（周一）下午 13:30-14:30

报告地点

VENUE

线下：同济大学土木大楼 A305

线上：腾讯会议 ID：925389373密码：786700

主持人

CHAIR

牛立志、孙婷婷

报告摘要

Abstract

In this talk, Dr. Kai Cheng will present a novel surrogate model, termed state space Kriging (S2K), for emulating high-dimensional nonlinear stochastic dynamical systems. S2K approximates nonlinear stochastic dynamical systems in their state space form, whereby every stochastic excitation enters the state space equation as a one-dimensional parameter. In this manner, the proposed surrogate model circumvents the “curse of dimensionality” that arises from the discretization of the stochastic excitation. For problems with partial observation of state variables, the time delay embedding technique is combined with S2K to complement the spatial state variables with time-delayed state variables. Model reduction technique is further employed to reduce computational complexity resulting from the high dimensionality of the state-space equation for systems with large degrees of freedom. Numerical examples show that the proposed S2K model yields stable and accurate predictions with only a few training time histories.

报告人简介

Speaker Bio

Dr. Kai Cheng is an Alexander von Humboldt Research Fellow at the Engineering Risk Analysis Group, Technical University of Munich (TUM). Before joining TUM, he was a Postdoctoral Researcher at the Department of Mathematics and Computer Science, University of Southern Denmark (2021–2023). He received his Ph.D. degree (2018–2021) and M.Sc. degree (2015–2018) in Flight Vehicle Design from Northwestern Polytechnical University, and his B.Sc. degree (2011–2015) in Mechanics from China University of Petroleum (East China). His research focuses on uncertainty quantification and reliability analysis of high-dimensional nonlinear stochastic dynamical systems. His main interests include stochastic dynamics, surrogate modeling and machine learning, model reduction of dynamical systems, rare-event probability estimation, and uncertainty quantification of engineering models.