Structural response self-prediction method based on long short-term memory neural network and compressive sensing

Jian Zhou; Hong-Wei Li; Yi-Qing Ni; Qiu-Sheng Li; You-Wu Wang

doi:10.1177/14759217251332002

Structural response self-prediction method based on long short-term memory neural network and compressive sensing

Jian Zhou
, Hong-Wei Li^*
, Yi-Qing Ni
, Qiu-Sheng Li
, You-Wu Wang

^*Corresponding author for this work

Department of Architecture and Civil Engineering

Research output: Journal Publications and Reviews › RGC 21 - Publication in refereed journal › peer-review

1 Citation (Scopus)

62 Downloads (CityUHK Scholars)

Abstract

The utilization of data-driven models, for example, the long short-term memory neural network (LSTM) model, has emerged as a potential structural response prediction approach for tackling the issues with the transmission and storage of long-term monitoring data. However, the prediction error of the LSTM models may increase significantly as the total prediction length increases, limiting the applicability of LSTM in response prediction. To tackle this challenge, this study proposes a hybrid prediction method, namely LSTM + compressive sensing (LSTM + CS), which combines LSTM and CS to improve prediction accuracy. CS embeds physical information into the prediction process, thereby the prediction divergence can be well mitigated. Simulated responses of a four-degree-of-freedom system and real-world responses of the Canton Tower are utilized for verification purposes, demonstrating that LSTM + CS can achieve a high prediction performance with limited data and minimal costs. The proposed method has the potential to be an effective response prediction tool to reduce the data storage requirement. © The Author(s) 2025.

Original language	English
Number of pages	14
Journal	Structural Health Monitoring
DOIs	https://doi.org/10.1177/14759217251332002
Publication status	Online published - 21 Apr 2025

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The work described in this article was supported by the Hong Kong Polytechnic University\u2019s Start-up Fund for RAPs under the Strategic Hiring Scheme (Grant No. P0046770). The authors also appreciate the funding support by the Innovation and Technology Commission of Hong Kong SAR Government to the Hong Kong Branch of National Engineering Research Center on Rail Transit Electrification and Automation (Grant No. K-BBY1).

Research Keywords

compressive sensing
long short-term memory network
machine learning
structural health monitoring
Structural response hybrid prediction

Publisher's Copyright Statement

COPYRIGHT TERMS OF DEPOSITED POSTPRINT FILE: The article is protected by copyright and reuse is restricted to non-commercial and no derivative uses. Users may also download and save a local copy of an article accessed in an institutional repository for the user's personal reference. For permission to reuse an article, please follow our Process for Requesting Permission. Zhou J, Li H-W, Ni Y-Q, Li Q-S, Wang Y-W, Structural response self-prediction method based on long short-term memory neural network and compressive sensing, Structural Health Monitoring, Ahead of print. Copyright © 2025 The Author(s). DOI: 10.1177/14759217251332002

Access Output

10.1177/14759217251332002

282813069.pdfPost-print, 1.22 MB

Other Access Options

Check CityUHK Library

Permanent Link

Right click to copy link

Cite this

@article{b8f6bdaa4fc34f49ab00922ef0f5b045,

title = "Structural response self-prediction method based on long short-term memory neural network and compressive sensing",

abstract = "The utilization of data-driven models, for example, the long short-term memory neural network (LSTM) model, has emerged as a potential structural response prediction approach for tackling the issues with the transmission and storage of long-term monitoring data. However, the prediction error of the LSTM models may increase significantly as the total prediction length increases, limiting the applicability of LSTM in response prediction. To tackle this challenge, this study proposes a hybrid prediction method, namely LSTM + compressive sensing (LSTM + CS), which combines LSTM and CS to improve prediction accuracy. CS embeds physical information into the prediction process, thereby the prediction divergence can be well mitigated. Simulated responses of a four-degree-of-freedom system and real-world responses of the Canton Tower are utilized for verification purposes, demonstrating that LSTM + CS can achieve a high prediction performance with limited data and minimal costs. The proposed method has the potential to be an effective response prediction tool to reduce the data storage requirement. {\textcopyright} The Author(s) 2025.",

keywords = "compressive sensing, long short-term memory network, machine learning, structural health monitoring, Structural response hybrid prediction",

author = "Jian Zhou and Hong-Wei Li and Yi-Qing Ni and Qiu-Sheng Li and You-Wu Wang",

year = "2025",

month = apr,

day = "21",

doi = "10.1177/14759217251332002",

language = "English",

journal = "Structural Health Monitoring",

issn = "1475-9217",

publisher = "Sage Publications",

}

TY - JOUR

T1 - Structural response self-prediction method based on long short-term memory neural network and compressive sensing

AU - Zhou, Jian

AU - Li, Hong-Wei

AU - Ni, Yi-Qing

AU - Li, Qiu-Sheng

AU - Wang, You-Wu

PY - 2025/4/21

Y1 - 2025/4/21

N2 - The utilization of data-driven models, for example, the long short-term memory neural network (LSTM) model, has emerged as a potential structural response prediction approach for tackling the issues with the transmission and storage of long-term monitoring data. However, the prediction error of the LSTM models may increase significantly as the total prediction length increases, limiting the applicability of LSTM in response prediction. To tackle this challenge, this study proposes a hybrid prediction method, namely LSTM + compressive sensing (LSTM + CS), which combines LSTM and CS to improve prediction accuracy. CS embeds physical information into the prediction process, thereby the prediction divergence can be well mitigated. Simulated responses of a four-degree-of-freedom system and real-world responses of the Canton Tower are utilized for verification purposes, demonstrating that LSTM + CS can achieve a high prediction performance with limited data and minimal costs. The proposed method has the potential to be an effective response prediction tool to reduce the data storage requirement. © The Author(s) 2025.

AB - The utilization of data-driven models, for example, the long short-term memory neural network (LSTM) model, has emerged as a potential structural response prediction approach for tackling the issues with the transmission and storage of long-term monitoring data. However, the prediction error of the LSTM models may increase significantly as the total prediction length increases, limiting the applicability of LSTM in response prediction. To tackle this challenge, this study proposes a hybrid prediction method, namely LSTM + compressive sensing (LSTM + CS), which combines LSTM and CS to improve prediction accuracy. CS embeds physical information into the prediction process, thereby the prediction divergence can be well mitigated. Simulated responses of a four-degree-of-freedom system and real-world responses of the Canton Tower are utilized for verification purposes, demonstrating that LSTM + CS can achieve a high prediction performance with limited data and minimal costs. The proposed method has the potential to be an effective response prediction tool to reduce the data storage requirement. © The Author(s) 2025.

KW - compressive sensing

KW - long short-term memory network

KW - machine learning

KW - structural health monitoring

KW - Structural response hybrid prediction

UR - https://www.scopus.com/pages/publications/105004300980

UR - https://www.scopus.com/record/pubmetrics.uri?eid=2-s2.0-105004300980&origin=recordpage

U2 - 10.1177/14759217251332002

DO - 10.1177/14759217251332002

M3 - RGC 21 - Publication in refereed journal

SN - 1475-9217

JO - Structural Health Monitoring

JF - Structural Health Monitoring

ER -

Structural response self-prediction method based on long short-term memory neural network and compressive sensing

Abstract

Funding

Research Keywords

Publisher's Copyright Statement

Access Output

Other Access Options

Permanent Link

Other Files and Links

Fingerprint

Cite this