Academy of Mathematics and Systems Science, CAS
Colloquia & Seminars

Speaker:	邹家辉, 首都经济贸易大学
Inviter:
Title:	Earthquake prediction using a three-stage one-dimensional CNN model
Language:	Chinese
Time & Venue:	2024.01.30 09:00-10:00 腾讯会议号：962 4818 2651 内部讨论
Abstract:	Abstract Earthquakes are severe natural disasters that often pose a significant threat to human life and property. It would bring great benefits to people if it is possible to predict the location, occurrence time and magnitude of an earthquake in advance. Earthquake prediction is also a vital issue in data science research. This paper aims to predict the occurring time, location and magnitude of the impending earthquakes in the upcoming 7 days. We develop a three-stage Convolutional Neural Network (CNN) to achieve this task for the southwest seismic zone of China (22? ～ 34?N, 98? ～ 107?E), based on the electromagnetic and geoacoustic signals recorded by AETA system. The training process takes place from January 1st, 2017, to April 30th, 2022. The testing process occurs from July 1st, 2022, to December 30th, 2022. Finally, our method has an accuracy of 60% in predicting earthquakes, and the false positive rate (FPR) is 37.5%.