Physics-encoded interpretable self-supervised learning for structural damage identification

Jiaxin Liu

doi:10.1016/j.engstruct.2025.121045

Physics-encoded interpretable self-supervised learning for structural damage identification

2025-08-05

0

PRE

AI

DOI：10.1016/j.engstruct.2025.121045

Original

Original request for help

Save

Abstract

En 中文

• Propose an interpretable neural network by embedding structural dynamics. • Train the deep learning model in a novel self-supervised learning way. • Raw response data are directly used without elaborate feature extraction. • The method is output-only and data-based.

Keywords:

interpretable neural network

structural dynamics

self-supervised learning

raw response data

output-only method

Physics-encoded interpretable self-supervised learning for structural damage identification

Abstract

Journal

Organization