基于小波域的FastICA算法的非常规油气藏地震资料去噪  

The application of Fast ICA algorithm based on wavelet domain to seismic data denoising of unconventional oil and gas reservoirs

作　　者：王顾希[1,2] 郭思[1,3] 郭科[1] 彭宇[3] 

机构地区：[1]成都理工大学,四川成都610059 [2]中国测试技术研究院,四川成都610021 [3]川庆钻探工程有限公司地质勘探开发研究院,四川成都610000

出　　处：《地质通报》2015年第7期1391-1399,共9页Geological Bulletin of China

基　　金：国家自然科学基金项目(批准号:41272363);2013高等学校博士学科点专项(编号:20135122110010)

摘　　要：近年来,随着鄂尔多斯盆地苏里格气田开发规模的不断扩大,全数字三维地震得到广泛应用,数字检波器的高灵敏性使得地震资料噪声极其发育。因此,全数字三维地震资料叠前去噪是资料处理的关键。针对全数字三维地震资料的特点,结合盲信号分离技术的最新发展,特别是神经网络技术的最新发展应用到研究中来,建立适合于理论地震记录的盲信号分离的算法模型,对已知反射地震数据实施盲分离技术,将地震信号变换到小波域中,并用Fast ICA算法进行盲分离去噪,然后将去噪后的信号从小波域变换到时间域信号。试验结果表明,该方法得到的去噪效果较时间域内直接去噪效果好。In recent years, with the continuous expansion and development of Sulige gas field in Ordos Basin, the digital 3D seismic data have been widely used, and digital geophone with high sensitivity that causes noises in seismic data is extremely developed.Therefore, the digital 3D seismic data prestack denoising is the key to data processing. In this paper, based on the 3D digital seismic data, in combination with the new development of the blind signal separation technology, especially the latest development and application of the neural network technology, the authors studied and established the blind signal separation algorithm model suitable for theoretical seismograms, performed blind source separation technology for known seismic reflection data, transformed the seismic signal into wavelet domain, used Fast ICA algorithm to conduct blind separation denoising, and then transformed the denoised signal from the wavelet domain to the time domain. The experimental results show that the denoising result of this method is better than the direct denoising result in the time domain.

关 键 词：地震资料 三维地震 非常规油气藏 小波变换 随机噪声 阈值函数 

分 类 号：P628[天文地球—地质矿产勘探]