地震波阻抗反演的ANNLOG技术及其应用效果
|
摘要
通过地震子波的多级分解和多级非线性变换，得到一种非线性地震褶积模型。将该模型与F－P模型人工神经网络理论相结合，可形成一套利用测井和地层约束的高分辨率地震波阻抗反演技术（ANNLOG）。其突出的特点是：多级非效性变换能使迭代反演快速收敛，并具有极高的纵向反演分辨率；用于存储多级地震子波的人工神经网络，可根据地震数据动力学特征在横向上的变化进行可靠的自适应外推反演，并在横向上保持纵向分辨率的连续性；采用地层约束反演，使ANNLOG技术适用于大断距断层、地层尖灭等复杂地质构造情况。
A nonlinear seisimc convolution model is constructed by performing the multi-stage decomposition and multistage nonlinear transform of seismic wavelets. Com-bining this model with the neural network that is based on neuron F-P functionmodel forms a ANNLOG technique for high-resolution wave impedance inversionunder the c0nstraints of logging and stratigraphic data. This technique involves fol-lowing essential points:. Multistage nonlinear transform causes fast iterative convergence and veryhigh resolution in vertical direction.. According to the lateral dynamic characteristic variation of seismic data,theneural networks storing multistage seismic wavelets (F-P model) enable reliableadaptive extrapo1ation inversion, and make continuous vertical resolution be stablein azimuthal direction.. Stratigraphic restrained inversion makes ANNLOG technique suitable to complexgeologic structures such as big throw faults,pinchouts and so on.
A nonlinear seisimc convolution model is constructed by performing the multi-stage decomposition and multistage nonlinear transform of seismic wavelets. Com-bining this model with the neural network that is based on neuron F-P functionmodel forms a ANNLOG technique for high-resolution wave impedance inversionunder the c0nstraints of logging and stratigraphic data. This technique involves fol-lowing essential points:. Multistage nonlinear transform causes fast iterative convergence and veryhigh resolution in vertical direction.. According to the lateral dynamic characteristic variation of seismic data,theneural networks storing multistage seismic wavelets (F-P model) enable reliableadaptive extrapo1ation inversion, and make continuous vertical resolution be stablein azimuthal direction.. Stratigraphic restrained inversion makes ANNLOG technique suitable to complexgeologic structures such as big throw faults,pinchouts and so on.
引文
1JeanBrac,Pierre Yves Kequirez et al.Inversion with a prior in formation:an approach to integrated strati-graphic interpretation,Soc. Expl. Geophys. Erpanded Abstracts of 58th SEG Mtg, 1988, 841～844
2Martinez R D et al.Complex reservoir characterization by multiparameter constrained Inversion,SEG/EAEG Research Workshop on Reservoir Geophysics,Callas,Texas, 1988
3 Dubrovsky Z M. Parametric CDP section Inversion technology PARM,Expanded Abstracts of 59th SEGMtg,1989,505～508
4Didier Carron E P. Well guided stratigraphic Inversion of borehole and surface sections,ErpandedAbstracts of 58th SEG Mtg,1988:837～840
5符力耘,牟永光.地震道非线性合成模型及地震储层参数反演.石油物探东部地区第七次学术研讨会论文集,1995,127～129
6符力耘.人工神经网络理论及其在地震信号处理中的应用,石油大学博士论文,1995
2Martinez R D et al.Complex reservoir characterization by multiparameter constrained Inversion,SEG/EAEG Research Workshop on Reservoir Geophysics,Callas,Texas, 1988
3 Dubrovsky Z M. Parametric CDP section Inversion technology PARM,Expanded Abstracts of 59th SEGMtg,1989,505～508
4Didier Carron E P. Well guided stratigraphic Inversion of borehole and surface sections,ErpandedAbstracts of 58th SEG Mtg,1988:837～840
5符力耘,牟永光.地震道非线性合成模型及地震储层参数反演.石油物探东部地区第七次学术研讨会论文集,1995,127～129
6符力耘.人工神经网络理论及其在地震信号处理中的应用,石油大学博士论文,1995
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