Seismic magnitude rapid estimation method based on deep learning feature fusion

Abstract

The invention discloses a seismic magnitude rapid estimation method based on deep learning feature fusion. The method comprises the following steps: (1) collecting a large number of seismic monitoringstation ground vibration waveform records and corresponding seismic catalogues; (2) calculating traditional parameter characteristics and statistical characteristics of a data sample; (3) training adepth model, and extracting image features and time sequence features in the data sample; and (4) fusing the features extracted from the data sample to form vector representation, training an XGBoostmodel to obtain a seismic magnitude rapid estimation model, and further utilizing the model to obtain a corresponding seismic magnitude according to seismic waveform record estimation. Research results of traditional seismology are utilized, and are combined with deep learning, a seismic magnitude rapid estimation model is learned by utilizing a large amount of seismic data, and the accuracy of seismic magnitude rapid estimation is improved.

Description

technical field [0001] The invention belongs to the technical field of earthquake early warning, and in particular relates to a method for rapidly estimating earthquake magnitude based on deep learning feature fusion. Background technique [0002] Over the years, geologists have used the continuous waveform data recorded by seismic monitoring stations to conduct a lot of work and research in the fields of earthquake early warning, earthquake quick report, and earthquake mechanism. Among them, tasks such as seismic event detection, automatic seismic phase picking, and rapid magnitude estimation are the focus and hotspots of related research. [0003] The task of rapid earthquake magnitude estimation is very important in earthquake early warning. If an accurate estimate of the magnitude of an earthquake can be given a few seconds before an earthquake occurs, it will be of great help to earthquake early warning and disaster relief work. , for more time. [0004] The research ...

AI Technical Summary

Problems solved by technology

However, the current research on the deep learning application of the seismic magnitude rapid estimation task is not very common, and the traditional method is still the main method, and there is currently a lack of research work on the integration of deep learning technology and traditional methods.

Images