Intelligent seismic data de-aliasing method and system based on U-Net network

Abstract

The invention relates to an intelligent seismic data de-aliasing method and system based on a U-Net network. The method comprises the following steps of (1) constructing the U-Net network f for seismic data de-aliasing; (2) obtaining a simulated data training pair, wherein the simulated data training pair comprises the simulated non-aliasing seismic data and the aliasing seismic data; (3) trainingthe U-Net network by taking the simulated aliasing seismic data as input and the non-aliasing seismic data as a training label to obtain a trained network parameter theta*; (4) on the basis of a transfer learning method, fine tuning the trained network parameter theta* by using part of actual aliasing seismic data containing labels to obtain an optimized network parameter; and (5) carrying out loop iteration on the seismic data to be processed by using the optimized U-Net network to obtain the separated seismic data. Compared with the prior art, the hypotheses of linearity, sparsity, low rankand the like of the data are avoided, the de-aliasing processing efficiency is high, the stability is good, and the precision is high.

Description

technical field [0001] The invention relates to a seismic data anti-aliasing method and system, in particular to an intelligent seismic data anti-aliasing method and system based on a U-Net network. Background technique [0002] Seismic data aliasing acquisition is one of the methods to effectively improve the efficiency of seismic data acquisition, but the existence of aliasing noise brings certain challenges to subsequent seismic data processing and migration imaging, so the anti-aliasing method has received extensive attention. [0003] Traditional anti-aliasing methods generally include filtering methods and inversion methods: filtering methods include median filtering, improved median filtering, FX deconvolution, etc.; inversion methods include threshold iteration based on sparse transformation, dictionary learning, etc. shrink method. Traditional anti-aliasing methods generally have certain prerequisites for seismic data, such as linearity, sparsity, or low rank; in a...

AI Technical Summary

Problems solved by technology

Traditional anti-aliasing methods generally have certain prerequisites for seismic data, such as linearity, sparsity, or low rank; in addition, as the scale of seismic data increases, the amount of calculation increases rapidly, and the optimization parameters corresponding to different data are different. Manual adjustment is required, which further increases the calculation time

Images