Time-varying wavelet extraction method based on local similarity and evaluation feedback

Abstract

The invention brings forward a time-varying wavelet extraction method based on local similarity and evaluation feedback, and belongs to the seismic prospecting signal processing field. The method is characterized by including the following steps: extracting the amplitude spectrum of a time-varying wavelet pointwise through a time frequency domain spectrum simulation method, adopting a bispectrum method to pre-estimate the phase of the wavelet and determining the phase searching scope, improving the precision of the estimated phase by combining with a phase optimization searching method based on local similarity, and finally obtaining a high-precision time-varying seismic wavelet. To quantitatively verify the wavelet extraction accuracy, the method carries out deconvolution processing for an earthquake record through the extracted wavelet, evaluates a deconvolution result according to evaluation principles, and thus indirectly evaluates the wavelet extraction accuracy. Meanwhile, the principle of Parsimony is combined with a singular value decomposition technology to solve the problem that the signal to noise ratio of the earthquake record is low, an evaluation SVD_P principle with a high noise-immune capability is constructed, and finally an entire time-varying wavelet extraction method and time-varying wavelet extraction accuracy evaluation method are formed. The synthetic non-stationary earthquake record and actual data processing on the basis of an optimization algorithm verify the feasibility and advantage of the method.

Description

Technical field: [0001] The invention belongs to the field of seismic exploration signal processing. Background technique: [0002] The research purpose of seismic data processing is mainly carried out around the three highs, that is, high signal-to-noise ratio, high resolution, and high fidelity. Basic processing techniques for seismic data include deconvolution, stacking, and migration. Stacking technology improves the signal-to-noise ratio of seismic data by suppressing noise, while using migration imaging technology to improve spatial resolution and fidelity is achieved through spatial homing of the interface and recovery of wave field characteristics. Deconvolution technology separates seismic wavelets and reflection coefficients by compressing wavelets to improve time resolution. As researchers continue to develop and improve random signal processing technology, oil exploration data processing technology is also developing in a more accurate and efficient direction. ...

AI Technical Summary

Problems solved by technology

Most of the above methods are based on the assumption of segmental stabilization, that is, after the non-stationary seismic records are divided into time windows, each segment is regarded as an approximately stationary seismic record to extract wavelets, but this method cannot well reflect the wavelet differences between adjacent intervals. changes, and is affected by the segment length, there must be some errors in the results

[0005] For the judgment of the accuracy of wavelet extraction in non-stationary seismic records, it is difficult to directly evaluate the accuracy whether it is from the wavelet waveform, energy or the properties of the wavelet itself.

Using unverified wavelets for subsequent deconvolution processing and wave impedance inversion may cause error amplification and artifacts, seriously affecting the accuracy of oil and gas resource exploration

Images