Applying Deep Learning to Continuous Bridge Deflection Detected by Fiber Optic Gyroscope for Damage Detection

Area of Science:

• Civil Engineering
• Structural Health Monitoring
• Artificial Intelligence

Background:

• Bridge structure damage detection is crucial for safety and requires improved accuracy and efficiency.
• Current methods face challenges in precise and timely identification of structural integrity issues.

Purpose of the Study:

• To develop an advanced method for continuous bridge deflection monitoring using fiber optic gyroscopes.
• To apply deep learning algorithms for accurate structural damage detection in bridges.

Main Methods:

• A scale-down bridge model was used to simulate three damage scenarios and an intact benchmark.
• A supervised learning model based on deep convolutional neural networks (CNNs) was developed and trained.
• Ten-fold cross-validation was employed for model training and performance evaluation.

Main Results:

• The proposed deep convolutional neural network model achieved a high accuracy of 96.9% in damage detection.
• The CNN model significantly outperformed traditional machine learning methods (random forest, SVM, KNN, decision tree).
• The model demonstrated a notable capability in differentiating damage from structurally symmetrical locations.

Conclusions:

• The integration of fiber optic gyroscope monitoring and deep learning offers a highly accurate and efficient approach to bridge damage detection.
• Deep convolutional neural networks provide a superior alternative to traditional machine learning for structural health monitoring of bridges.
• The developed method shows promise for real-world applications in ensuring bridge safety and structural integrity.