A transformer-based real-time earthquake detection framework in heterogeneous environments

Area of Science:

• Seismology
• Artificial Intelligence
• Internet of Things

Background:

• Deep learning methods show superior efficacy in earthquake detection compared to conventional approaches.
• Deploying deep learning in heterogeneous environments presents challenges due to data variations and diverse evaluation metrics.
• Current models often prioritize S-wave detection, overlooking the critical early identification of P-waves.

Purpose of the Study:

• To introduce TFEQ, a novel transformer-based model for real-time earthquake detection.
• To enable concurrent analysis of both P and S waves for improved seismic event identification.
• To validate the model's effectiveness and applicability in diverse Internet of Things (IoT) environments.

Main Methods:

• Development of TFEQ, a transformer-based deep learning architecture.
• Concurrent analysis of P-wave and S-wave seismic data.
• Case studies utilizing MEMS sensor data from the CrowdQuake initiative.

Main Results:

• TFEQ demonstrates effective real-time earthquake detection capabilities.
• The model concurrently analyzes P and S waves across different data domains.
• Case studies confirm TFEQ's reliability and generalization across diverse datasets.

Conclusions:

• TFEQ offers a robust solution for real-time earthquake detection in heterogeneous IoT settings.
• Concurrent P and S wave analysis enhances seismic event identification accuracy.
• The model shows significant potential for improving seismological research and disaster preparedness.