College - Author 1
College of Engineering
Department - Author 1
Computer Engineering Department
Degree Name - Author 1
BS in Computer Engineering
Date
6-2026
Primary Advisor
Xiao-Hua Yu, College of Engineering, Electrical Engineering Department
Abstract/Summary
Early earthquake warning (EEW) systems are critical in warning citizens in earthquake prone areas of impending seismic activity, ensuring safety and minimizing damage to property. To develop an early earthquake warning system, the categories of a seismic event need to be predicted in real-time, and this task can be done using deep learning techniques such as convolutional neural networks. Four convolutional neural networks were trained to identify earthquake magnitude ranges, depth ranges, location clusters, and origin time clusters individually along with one neural network that can make all the classifications using one mode. The individual models performed with accuracies of 88.96%, 74.5%, 95.35%, and 91.95% for magnitude, depth, location, and origin time respectively. The model to predict all categories at once has similar performance and all models can make prediction decisions within 1 millisecond which is good for fast warning times.
URL: https://digitalcommons.calpoly.edu/cpesp/355