摘要
利用人工神经网络的BP算法，对二维叠后地震剖面的地层岩性进行估计。首先选择井眼处地层的孔隙率、泥质含量作为网络的预测对象，将井旁地震道的地震属性参数（瞬时频率、相对波阻抗等）作为输入样本，求取各层节点的权值。当井旁地震道训练结束后，再利用已确定好的孔隙率网络和泥质含量网络外推，计算其余地震道相应地层的孔隙率和泥质含量剖面。经数值计算结果表明，选择径向基函数作为隐层的激励函数，可以得到较好的样本拟合效果。
Lithology of 2-D post-stack seismic section can be predicted with the use of BPalgorithm of neural network. First, the porosity and shale content of formationsaround the chosen borehole are taken as prediction target in the network, and theseismic parameters (such as instantaneous frequency and re1ative wave impedance)of borehole-side trace are used as input samples to solve for node-point weight values of each layer. After the training of borehole-side seismic trace is finished, thedetermined porosity network and shale content network are extrapolated to compute the porosity and shale content of the corresponding formation of other seismictraces.Trial numerical computation indicates that taking radial basic function as exciting function of a hidden layer brings good sample fitting effect.
出处
《石油地球物理勘探》
EI
CSCD
北大核心
1996年第3期400-409,共10页
Oil Geophysical Prospecting
关键词
神经网络
BP算法
岩性
地震勘探
地层
地震剖面
neural network,BP algorithm,radial basic function, instantaneous frequency,wave impedance,formation,porosity,seismic section
