An extraction method of seismic background noise dispersion curve based on deep learning

Abstract

The invention discloses a method for extracting the dispersion curve of seismic background noise based on deep learning. First, the manual extraction of dispersion curve data is collected, and the collected data is divided into a training set, a verification set and a test set; a U -Net simplified neural network, and use the training set data to train the neural network; input the group velocity-periodogram and phase velocity-periodogram in the seismic background noise to be extracted into the trained neural network, and obtain the neural network prediction Group velocity and phase velocity energy diagram; combine the group velocity-periodogram, phase velocity-periodogram and the group velocity and phase velocity energy diagram predicted by the neural network to initially extract the dispersion curve; use the prior information of multiple dispersion curves to Constrains the extracted dispersion curve. The above method can quickly and accurately extract the group velocity and phase velocity dispersion curves of seismic background noise automatically, and then invert the underground structure, saving a lot of labor time and cost.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to a method for extracting the dispersion curve of seismic background noise based on deep learning. Background technique [0002] Seismic background noise imaging is one of the commonly used methods to study different deep underground structures in recent years. For example, using the continuous waveform data recorded by seismic stations, the surface wave Green's function between stations can be extracted through the cross-correlation and superposition process of stations. By summarizing the data processing process of seismic background noise, including single data preparation, background noise cross-correlation, segment superposition, dispersion curve extraction and quality control of dispersion data, and using the extracted surface wave dispersion based traveltime imaging It is used to invert the subsurface structure; based on the image analysis technology, the cr...

AI Technical Summary

Problems solved by technology

By summarizing the data processing process of seismic background noise, including single data preparation, background noise cross-correlation, segment superposition, dispersion curve extraction and quality control of dispersion data, and using the extracted surface wave dispersion based on traveltime imaging It is used to invert the subsurface structure; based on the image analysis technology, the cross-correlation velocity-period (c-T) diagram is calculated, and then the surface wave dispersion curve is extracted from it. This method has been widely used to extract the cross-correlation function from the background noise Fundamental surface wave dispersion; the dispersion curve can also be automatically extracted by tracing the curves connected by the amplitude maxima in the c-T diagram, but this method has the problem that it is difficult to select the phase velocity branch caused by the double station method

[0003] In the prior art, in order to obtain a reliable dispersion curve, it is usually necessary to manually select control points in the c-T diagram to help the program track the correct dispersion curve. In order to obtain a more detailed underground structure, dense seismic stations need to be arranged in a research area The number of stations is increasing rapidly, because this method extracts the dispersion data between any two stations, which greatly increases the number of dispersion curves that need to be manually extracted, and consumes a lot of human resources, so it is necessary to develop a An Accurate and Reliable New Method for Extracting Dispersion Curves of Seismic Background Noise Without Any Human Interaction

Images