Related Experiment Video
Updated: Aug 19, 2025
06:49
Evaluation of a Smartphone-based Human Activity Recognition System in a Daily Living Environment
Published on: December 11, 2015
9.0K
Earthquake Event Recognition on Smartphones Based on Neural Network Models
Meirong Chen1, Chaoyong Peng1,2, Zhenpeng Cheng1
1Institute of Geophysics, China Earthquake Administration, Beijing 100081, China.
Sensors (Basel, Switzerland)
|November 26, 2022
Summary
Smartphone sensors can improve earthquake early warning (EEW) systems by using neural networks to accurately detect seismic events, reducing costs and server load. This enhances EEW accuracy and lead time.
Area of Science:
- Geophysics
- Seismology
- Mobile Computing
Background:
- Traditional earthquake early warning (EEW) systems are costly.
- Smartphone sensors offer a potential low-cost alternative for EEW.
- Distinguishing seismic events from human activity on mobile phones is challenging.
Purpose of the Study:
- Investigate neural network models for improved seismic event detection on smartphones.
- Enhance the accuracy of earthquake recognition using mobile phone data.
- Reduce the costs associated with traditional EEW systems.
Main Methods:
- Collected three-component acceleration data from human activities (walking, running, cycling) using a mobile app.
- Combined human activity data with seismic event records and mobile phone noise.
- Trained and tested fully connected and convolutional neural network models.
Main Results:
- Neural network models achieved over 98% accuracy in seismic event detection.
- Precision for seismic events and recall for non-earthquake events reached 99%.
- Demonstrated significant enhancement in seismic event recognition accuracy on smartphones.
Conclusions:
- Neural networks effectively improve seismic event recognition on smartphones.
- This approach significantly reduces data transmission and server load for EEW systems.
- Smartphone-based EEW systems can offer increased lead time and reduced costs.
