Early damage detection in bridges using a variational autoencoder-based hybrid unsupervised learning framework
These videos have been matched automatically. Contact us if they are not relevant.
These videos have been matched automatically. Contact us if they are not relevant.
View abstract on PubMed
Summary
This summary is machine-generated.This study enhances bridge structural health monitoring (SHM) using hybrid unsupervised machine learning (HUML) with variational autoencoders (VAEs). VAE-OCSVM integration offers superior early damage detection, even with missing data and environmental variability.
Area Of Science
- Civil Engineering
- Machine Learning
- Structural Health Monitoring
Background
- Structural health monitoring (SHM) is crucial for bridge safety but faces challenges like missing data and environmental variability.
- Existing methods struggle with early damage detection under these conditions.
Purpose Of The Study
- To evaluate hybrid unsupervised machine learning (HUML) frameworks, particularly variational autoencoders (VAEs), for early bridge damage detection.
- To compare the performance of six VAE-based models under severe environmental and operational variability (EOV) and missing data.
Main Methods
- Employed a four-step framework: initial data analysis (IDA), VAE-based latent representation, HUML for damage indicators (DIs), and extreme value theory (EVT) for thresholding.
- Evaluated VAE, VAE-OCSVM, VAE-IF, VAE-LOF, VAE-DBSCAN, and VAE-MSD on the Z24 Bridge dataset.
- Assessed performance based on decision errors, robustness against EOV, and threshold-free detection.
Main Results
- Integrating VAEs with anomaly detectors significantly improved damage detection.
- VAE-OCSVM demonstrated the highest precision, recall, specificity, and robustness against EOV.
- VAE-IF and VAE-DBSCAN showed comparatively weaker performance.
Conclusions
- Hybrid unsupervised machine learning frameworks, especially VAE-OCSVM, are effective for early damage detection in bridges.
- The proposed framework successfully mitigates challenges of missing data and EOV in structural health monitoring.
These videos have been matched automatically. Contact us if they are not relevant.
These videos have been matched automatically. Contact us if they are not relevant.
Related Concept Videos
The utilization of strain gauges as transducers for converting mechanical strain into electrical signals is a common practice in various engineering applications. These strain gauges are frequently integrated into Wheatstone bridge circuits to accurately measure parameters such as force or pressure. Within this context, each element within the circuit exhibits a resistance that undergoes subtle variations when subjected to mechanical strain. The primary objective is to convert minuscule...
Joints, also known as articulations, are classified based on their structural characteristics, i.e., based on whether the articulating surfaces of the adjacent bones are directly connected by fibrous connective tissue or cartilage, or whether the articulating surfaces contact each other within a fluid-filled joint cavity. These differences serve to divide the joints of the body into three structural classifications.
A fibrous joint is where the adjacent bones are united by fibrous connective...