Deep Learning-Based Real-Time Auto Classification of Smartphone Measured Bridge Vibration Data.

Area of Science:

• Engineering
• Computer Science
• Data Science

Background:

• Structural health monitoring (SHM) of bridges is crucial for public safety.
• Traditional SHM systems can be costly and complex.
• Mobile platforms offer potential for accessible and real-time data acquisition.

Purpose of the Study:

• To develop a real-time bridge vibration classification model for mobile platforms.
• To create a customizable convolution neural network (CNN) framework for vibration analysis.
• To enable accurate structural condition assessment using smartphone sensors.

Main Methods:

• Utilized a simple and customizable CNN framework for model training.
• Employed multichannel time-series signals from smartphone accelerometers as input.
• Trained and validated the model using long-term bridge monitoring data.

Main Results:

• Achieved accurate real-time classification of bridge vibration categories.
• Demonstrated the practical feasibility of a smartphone-based SHM system.
• Developed an iOS application for on-device vibration classification.

Conclusions:

• Smartphone-based systems can provide low-latency, high-accuracy structural condition monitoring.
• The proposed CNN framework effectively classifies bridge vibrations.
• This approach eliminates data processing bottlenecks, promoting stable SHM.