【摘要】 开展了微电阻率扫描成像的裂缝识别、参数提取以及裂缝参数的地震高分辨率反演研究。在对微电阻率扫描成像测井数据处理和图像生成的基础上,计算了裂缝孔隙度、裂缝密度、裂缝长度、平均裂缝宽度、裂缝水动力宽度等5个裂缝参数,定量描述了裂缝的发育状况。以测井计算的裂缝参数为学习样本,开展了高分辨率非线性地震反演(AN-FIS),该技术将模糊技术与神经网络技术相结合,集模糊逻辑推理的结构性知识表达能力与神经网络的自学能力于一体。ANFIS采用(GD+LSE)混合算法自适应地进行调节,该算法减少了原始纯反向传播算法的收缩空间的维数,因而收敛速度非常快。在实际反演中,由井点出发构筑测井信息与井旁地震道的非线性映射关系,根据地下介质在横向上的变化特征来更新非线性映射关系,并在地质约束下进行自适应外推,实现高分辨率反演,获得了高分辨率的裂缝参数剖面,反演的裂缝参数与成像测井裂缝参数吻合较好。更多还原

【Abstract】 Studied are fracture identification and parameters extraction from micro resistivity images as well as seismic high resolution inversion of fracture parameters.Based on the log data processing and image generation of micro resistivity images,calculated are fracture porosity,fracture density,fracture length,average fracture width and fracture hydrodynamic width,and then quantificationally described are the development status of the fractures.With fracture parameters calculated from log data as learning samples,we studied the high resolution nonlinear seismic inversion method on the basis of ANFIS(Adaptive Neural Fuzzy Inferense System).The method combines fuzzy with neural network technique,and also composites the high express capability of fuzzy logic with high self-study capability of neural network as a whole.ANFIS includes twoparts(i.e.,alterable presupposition and conclusion) and mixes arithmetic(GD+LSE) to adapt by itself and reduce the shrinking dimension number of original reverse arithmetic.Therefore,the convergence speed is very fast.In practical inversion,we constructed the nonlinear map relationship between logging information and seismic channels beside the borehole.According to the transverse variation character of underground media,it updates the nonlinear map relationship,does self-adapted extrapolation by the geology constraint and achieves the high resolution inversion,thereby acquiring high resolution profile of the fracture parameters.The fracture parameters from inversion are consistent with that from imaging logging.更多还原