A condition diagnosis method for subway track structures employing distributed optical fiber sensing

Area of Science:

• Civil Engineering
• Geotechnical Engineering
• Sensor Technology

Background:

• Urban rail transit systems face increasing risks of track structure damage due to high-load operations.
• Traditional detection methods for subway tracks are limited in coverage and real-time performance, compromising operational security.
• Ensuring the safety and integrity of subway track structures is critical for reliable urban transportation.

Purpose of the Study:

• To propose a novel method for diagnosing subway track structure states using distributed fiber sensing.
• To enhance the accuracy and efficiency of detecting track anomalies compared to traditional methods.
• To provide reliable technical support for the intelligent monitoring and safety of subway track structures.

Main Methods:

• Developed a correlation model for strain monitoring data using an optimal space window to reduce computational complexity.
• Constructed a deep generative adversarial network (GAN) model incorporating residual learning for data analysis.
• Utilized spatial correlation analysis of symmetric measuring points and Mahalanobis distance of predicted residuals for diagnosis.

Main Results:

• The proposed method effectively eliminates periodic interferences like temperature variations.
• Accurate detection of local strain anomalies was achieved with a positioning error less than the measuring point interval (20 cm).
• The deep learning model demonstrated robust performance in identifying the orbital structure state.

Conclusions:

• The distributed fiber sensing and deep learning approach offers a significant advancement in subway track structure monitoring.
• This method provides a reliable and precise tool for ensuring the operational safety of urban rail transit.
• The findings support the implementation of intelligent monitoring systems for critical infrastructure like subway tracks.