基于L1范数的全变分地震信号反褶积优化算法

(辽河油田公司勘探开发研究院,辽宁 盘锦 124010)

地震信号; 反褶积; 去噪; L1范数; 全变分理论; 交替方向乘子法

Deconvolution Optimization Algorithm of Seismic Signals Based on L1 Norm of Total Variation
LU Mingde

(Exploration and Development Research Institute of Liaohe Oilfield Company,Panjin 124010,Liaoning,China)

seismic signals; deconvolution; denoising; L1 norm; the Total Variation theory; the alternating direction method of multipliers *

DOI: 10.20015/j.cnki.ISSN1000-0666.2023.0020

备注

基于借范数最优化的思想,提出一种基于L1范数的全变分地震信号反褶积优化算法。该算法基于L1范数全变分理论构建地震信号重建模型,同时将其转化为符合迭代与交替最小化的求解形式,通过交替方向乘子法设计地震信号的反褶积优化算法。该算法无需考虑反褶积使用的限制条件,可以在含噪声的情况下有效恢复地震信号,同时提高地震信号的分辨率和信噪比。使用该算法对合成信号和野外采集地震数据进行实验,结果表明:该算法提高了子波的主频,拓宽了有效频带,即使在信号受到较重噪声污染时,也可以获得较好的处理结果。
Deconvolution is an effective method to improve the resolution of seismic signals.Due to the influence of noises,the stability of the sub-waves obtained by many deconvolution methods is poor,and the reflection coefficients are mixed with the sub-waves more seriously.In the light of the idea of norm optimization,a deconvolution optimization algorithm of seismic signals based on L1 norm of total variation is proposed in this paper.Based on the L1 norm of total variational theory,the seismic signal reconstruction model is constructed and transformed into a solution form conforming to iterative and alternating minimization.Then,the deconvolution optimization algorithm of seismic signals is designed by alternating direction multiplier method.The proposed algorithm does not need to consider the restrictions of deconvolution.It can effectively recover the seismic signals with noise,and improve the resolution and signal-to-noise ratio(SNR)of the seismic signals.Experiments on synthetic signals and field seismic data show that the algorithm improves the dominant frequency of sub-waves and widens the effective frequency band.It also performs well even if the signals are polluted by serious noises.
·