Aftershock Predict based on Convolution Neural Networks

Abstract

Earthquake prediction is a difficult task Constrained within a certain spatiotemporal range earthquakes are only a probability event In a large area predicting earthquakes based on geographical events that have already occurred is reliable Predicting the duration of aftershocks under the condition that a major earthquake has already occurred is the research content of this article Extract 6 features from seismic phase data to predict the aftershock period We constructed a convolutional neural network model sorted out 855 data from 1351 data and trained the network The accuracy of training verification reaches 90 and the accuracy of testing reaches 100 After further refinement this model can be used to predict the duration of aftershocks in earthquakes Provide data guidance for earthquake rescue