工程可靠性与随机力学国际联合研究中心

2025年第16期（总第105期）学术报告

工程力学研究中心第59期学术报告

土木工程学院院级高等讲堂

报告主题

TOPIC

工业结构设计与维护中的不确定性量化

Uncertainty quantification in industrial structural design and maintenance

报告人

SPEAKER

Dr Abhishek Kundu

MSc, PhD, FHEA, MRAeS, MASME, School of Engineering, Cardiff University, UK

报告时间

TIME

2025年10月17日（周五）下午15:30-16:30

报告地点

VENUE

同济大学土木大楼 A305

主持人

CHAIR

陈建兵教授、彭勇波教授

联系人：牛立志

报告摘要

Abstract

Uncertainty quantification (UQ) is a pervasive aspect of industrial structural applications, encompassing design, optimisation, monitoring, and prognostics. This presentation will provide an overview of a selection of research projects that employ a Bayesian approach to modelling, robust design under uncertainty, and degradation assessment for safety-critical industrial structures.

Robust early-stage industrial design of complex structures, such as aircraft wings, presents significant challenges due to design immaturity, a wide range of loading and operating conditions, and potential variability during manufacturing and assembly. The study presented herein aims to develop an approach for exploring the design space with a data-driven surrogate model that maps design parameters to a high-dimensional vector-valued output describing the response of quantities at locations under a range of operating scenarios. The optimal design of structural parameters under uncertainty is undertaken using a Bayesian inversion to identify probabilistic margins on design parameters in the presence of stochastic variability. The proposed approach is guided by a novel confidence-based criterion of meeting designer-specified performance thresholds. The results demonstrate the accuracy and computational savings of complex industrial design, and facilitates the expansion of the design space for complex industrial design workflows.

Continuous monitoring of safety-critical structures ensures reliable performance, maintenance, and prognostics. Acousto-ultrasonic signals monitor and interrogate these structures for concealed damage. Probabilistic characterisation of elastic properties reconstructs dispersive wave modes using Bayesian inversion, quantifying localised structural degradation. We explore minimal signal acquisition hardware footprint by integrating edge computing and physics-informed machine learning techniques for real-time structural monitoring. This approach is conceptualised as a Cyberphysical Structural Health Monitoring (CyberSHM) system, an automated monitoring framework integrated with the internet and collaborating with human end-users. The study utilises carbon-fibre composite panels with stiffeners as a test bench, subjecting them to impact and fatigue loading and monitoring with a

CyberSHM system, demonstrating its effectiveness, challenges, and a step towards realising the futuristic vision of automated continuous monitoring systems.

报告人简介

Speaker Bio

Dr. Abhishek Kundu is a Senior Lecturer at the School of Engineering at Cardiff University and the co-founder of the physics-informed digital twin laboratory at the Cardiff School of Engineering. He is an associate editor of ASME’s Journal of Non-destructive Testing. His research interests encompass uncertainty quantification, structural health monitoring (SHM) employing acoustic-ultrasonic techniques, noise and vibration control, Bayesian identification, cyberphysical systems, and digital twins. He obtained his PhD from Swansea University as a Zienkiewicz scholar and has authored over 60 scientific publications. His research has been recognised with esteemed awards, including the Harry Hilton Structures paper award at SciTech and the best paper award at EWSHM. He has participated in several academic and industrial projects including his recent EPSRC-funded project on CyberSHM for structural health monitoring as the Principle Investigator, his recent appointment as an Industrial Fellow with Airbus. He has been elected as a member of the Royal Aeronautical Society, UK.