Multi-wave adaptive subtraction method based on prediction feature extraction

Abstract

The invention provides a multi-wave adaptive subtraction method based on prediction feature extraction. The multi-wave adaptive subtraction method based on prediction feature extraction comprises thefollowing steps: 1, setting a variable initial value; 2, constructing a convolutional neural network; 3, training the convolutional neural network; 4, utilizing the trained convolutional neural network to extract and predict multiple wave characteristics; 5, constructing a mathematical model and an optimization problem of multi-wave adaptive subtraction by utilizing original data and predicting the multiple wave characteristics; 6, solving a matched filter of multi-wave adaptive subtraction; and 7, estimating a primary wave. By using the method for carrying out multi-wave adaptive subtractionby utilizing the predicted multiple wave characteristics, protection of the primary wave and separation of the multiple waves can be better balanced, and the residual multiple waves are effectively reduced while the primary wave is protected.

Description

technical field [0001] The invention relates to the field of oil and gas exploration seismic data processing, in particular to a multiple wave self-adaptive subtraction method based on prediction feature extraction. Background technique [0002] At present, seismic exploration pays more attention to complex structural oil and gas blocks such as deep oil and gas layers, and puts forward higher requirements for seismic imaging and inversion accuracy. The separation effect of primary wave and multiple wave has an important influence on the imaging and inversion of primary wave or multiple wave. The in-depth study of the separation method of the two is the frontier topic of seismic exploration. [0003] In the industry, predictive subtraction methods, such as SRME (Surface Related Multiple Elimination) method, are usually used to separate primary waves and multiple waves. Adaptive subtraction is the key to the separation of the two. How to effectively balance the protection of p...

AI Technical Summary

Problems solved by technology

Filters estimated by linear regression analysis methods are often unable to effectively express these complex differences

Especially in the exploration of complex geological structures, the primary wave and the multiple wave will have obvious intersecting or overlapping phenomena. At this time, the direct matching method in the linear regression analysis method is prone to under-fitting, resulting in residual multiple waves. Or produce overfitting and damage the primary wave

Images