Automated real-time cable tension warning framework using long-term monitoring data and a convolutional neural network
Authors:
- 26 May 2026
- journal article
- Published by SAGE Publications in Structural Health Monitoring
Abstract
Stay cables are critical load-bearing components in cable-stayed bridges, and their tension variations serve as direct indicators of structural safety and serviceability. While vibration-based monitoring has enabled continuous tension estimation, the reliable transformation of such measurements into real-time warning information remains challenging. This study presents an automated real-time cable tension warning framework that integrates ambient vibration measurements with a convolutional neural network (CNN). A statistical detection algorithm was first developed by defining thresholds from year-long tension histories and applying a three-segment strategy to suppress scattered values. Using long-term monitoring data from representative cables, this method successfully identified abnormal tension variations and quantified the minimum detectable changes and detection times. The proposed framework emphasizes warning logic and long-term operational performance by explicitly evaluating false-alarm behavior, detectability limits, and detection latency under unfavorable conditions. To overcome the inherent limitations of rule-based thresholds, the anomaly detection task was reformulated as a CNN-based image classification problem. By encoding tension data into one-dimensional images and categorizing them into low, normal, and high states, the CNN consistently achieved 100% accuracy in distinguishing normal from abnormal conditions. The contributions of this work are threefold: (1) development of a statistical threshold-based detection algorithm validated with 1 year of field data; (2) reformulation of anomaly detection as a CNN-based classification task; and (3) demonstration of perfect classification accuracy through extensive testing. These results establish a practical, adaptive framework for automated cable tension warning, extending conventional monitoring into proactive decision-making support for bridge management.Keywords
Funding Information
- National Science and Technology Council of Republic of China (MOST1102221E224007MY3)
- National Science and Technology Council of Republic of China (MOST1102221E224008)
- National Science and Technology Council of Republic of China (MOST1112221E224010)
- National Science and Technology Council of Republic of China (NSTC1132221E224024MY3)
This publication has 90 references indexed in Scilit:
- Automatic identification of structural modal parameters based on density peaks clustering algorithmStructural Control and Health Monitoring, 2022
- Automated operational modal analysis based on long‐term records: A case study of Milad Tower structural health monitoringStructural Control and Health Monitoring, 2022
- Stay cable tension estimation of cable‐stayed bridge under limited information on cable properties using artificial neural networksStructural Control and Health Monitoring, 2022
- Damage localization method for building structures based on the interrelation of dynamic displacement measurements using convolutional neural networkStructural Control and Health Monitoring, 2020
- Online early damage detection and localisation using multivariate data analysis: Application to a cable‐stayed bridgeStructural Control and Health Monitoring, 2019
- Automated modal identification using principal component and cluster analysis: Application to a long‐span cable‐stayed bridgeStructural Control and Health Monitoring, 2019
- Vibration‐based structural condition assessment using convolution neural networksStructural Control and Health Monitoring, 2018
- Convolutional neural network-based data anomaly detection method using multiple information for structural health monitoringStructural Control and Health Monitoring, 2018
- Automated long-term dynamic monitoring using hierarchical clustering and adaptive modal tracking: validation and applicationsJournal of Civil Structural Health Monitoring, 2018
- Application of stochastic subspace identification for stay cables with an alternative stabilization diagram and hierarchical sifting processStructural Control and Health Monitoring, 2016