论文部分内容阅读
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。
By seismic multi-level wavelet decomposition and multi-level nonlinear transformation, a nonlinear seismic convolution model is obtained. Combining this model with the F-P model artificial neural network theory, a set of high-resolution seismic impedance inversion techniques (ANNLOG) using well logging and formation constraints can be formed. Its prominent features are: multi-level ineffective transform can quickly converge iterative inversion and has a high vertical inversion resolution; artificial neural network for storing multi-stage seismic wavelet can be based on seismic data dynamics Features in the lateral changes of reliable adaptive extrapolation of inversion, and to maintain the vertical resolution of the continuity in the horizontal direction; using stratigraphic constrained inversion, the ANNLOG technology is suitable for large fault faults, strata and other complex geology Construction situation.