Temperature Effects Removal from Non-Stationary Bridge-Vehicle Interaction Signals for ML Damage Detection

Area of Science:

• Structural Engineering
• Artificial Intelligence
• Transportation Infrastructure

Background:

• Bridges are critical infrastructure requiring continuous safety monitoring.
• Traffic and environmental factors complicate damage detection in bridges.
• Existing methods struggle with combined operational and environmental variability.

Purpose of the Study:

• To propose and test a methodology for detecting and localizing bridge damage.
• To address challenges posed by traffic and temperature variability.
• To apply unsupervised machine learning for structural health monitoring.

Main Methods:

• Utilized principal component analysis for temperature effect removal from vibration data.
• Employed an unsupervised machine learning algorithm for damage detection and localization.
• Validated the methodology using a numerical bridge benchmark with simulated traffic loads and varying temperatures.

Main Results:

• The proposed machine learning approach effectively detects and localizes damage under operational and environmental variability.
• Temperature removal using PCA was successful in analyzing forced vibrations.
• The study demonstrates the promise of AI in complex bridge health monitoring.

Conclusions:

• Machine learning offers a promising solution for intricate bridge damage detection problems.
• The methodology shows potential for enhancing bridge safety and reliability.
• Future work will focus on validation with real-world data and more complex scenarios.