应用神经网络技术提高地震资料分辨率
|
摘要
通过神经网络技术以井数据和井旁地震数据作为输入，学习它们之间的褶积关系，利用这种关系可以求出地震子波通过求取子波的逆即得到反子波，使建立在褶积模型基础上的地震数据可以与反子波褶积得到波形压缩提高资料的分辨率。该方法和传统的反褶积方法的区别在于对子波没有最小相位假设，而地震子波一般是混合相位的．同时有井数据的参与使得对目的层的子注求取更为准确，通过反褶积使地震波形得到很好的压缩，提高了分辨率
To get the high resolution seismic data, the important technique is deconvolution. In general the wavelet is unknown, and seismic traces deconvoluted supposing the wavelet is minimum phase. We practice a new way to perform deconvolution. At first we obtain the wavelet of seismic traces by neural network trained with log reflectivity and trace near well. Then we can get the inverse wavelet that may be called deconvolution operator, and the operator acts all seismic records. At last we obtain high resolution profile.
To get the high resolution seismic data, the important technique is deconvolution. In general the wavelet is unknown, and seismic traces deconvoluted supposing the wavelet is minimum phase. We practice a new way to perform deconvolution. At first we obtain the wavelet of seismic traces by neural network trained with log reflectivity and trace near well. Then we can get the inverse wavelet that may be called deconvolution operator, and the operator acts all seismic records. At last we obtain high resolution profile.
引文
1黄绪德编著反相积与地震道及演北京:石油工业出版社,1992
2刘争平。人工神经网络在测井-地震资料联合反演中的应用研究:[学位论文]成都:成都理工学院地球物理勘探系,1997
3马在田,曹景忠,王家林,等计算地球物理学概论上海:同济大学出版社,1997
2刘争平。人工神经网络在测井-地震资料联合反演中的应用研究:[学位论文]成都:成都理工学院地球物理勘探系,1997
3马在田,曹景忠,王家林,等计算地球物理学概论上海:同济大学出版社,1997
版权所有：© 2023 中国地质图书馆 中国地质调查局地学文献中心