【作者】 曹自强；

【导师】 宋维琪；

【作者基本信息】 中国石油大学 ， 地球探测与信息技术， 2007， 硕士

【摘要】 储层预测技术是现代油藏管理和油藏描述的核心技术,地震资料、测井资料以及地质信息在储层预测中的综合利用为油气田勘探开发提供了良好技术保障。本文首先通过交替投影神经网络及扩展交替投影神经网络基本理论的概述,对交替投影神经网络及域扩展后的交替投影神经网络(扩展交替投影神经网络)收敛性理论进行描述和收敛性仿真。并应用扩展交替投影理论进行信号分离仿真计算。针对地震信号的特点,将扩展交替投影神经网络算法引入到地震领域,对模型和实际资料进行弱信号分离试算,在弱信号分离算法的基础上,结合滤波方法对特殊地质体进行弱信号加强算法研究,突出地下特殊岩体弱的反射特征。根据东营南坡高8地区沙四段上亚段滩坝砂储层地震响应被强背景信息淹没的特点,将算法应用到识别滩坝砂储层中,使反映砂体(组)形态的弱信号进行相对加强,突出砂体(组)在地震剖面上的展布特征。其次,对测井曲线采用信息重构的方法,综合自然电位曲线、自然伽玛曲线和声波时差曲线,依据输出信噪比最大的原则进行曲线重构处理。储层预测的最主要技术是波阻抗反演技术。最后,综合应用上述方法对地震和测井资料进行处理,应用Jason反演软件的约束稀疏脉冲反演模块,进行波阻抗反演。通过对处理前后反演结果的对比,得出经过资料处理后的反演结果可以较好的展示储层波阻抗差异特征。更多还原

【Abstract】 Reservoir prediction technique has become a core technique in reservoirmanagement and characterization. The information of seismic data, well logand geology information reasonably utilized has offered approved techniqueguarantee for exploration of oil and gas field. First of all, the paper gives asystematicsummaryofthebasictheoryof APNN andEAPNN, andemulatesconvergence for APNN and EAPNN. On the basic of EAPNN, it makes anexample of signal separation and imitation including no noise and containingnoise. In view of the seismic signal feature, we draw EAPNN method intoseismic domain, trying to separate weak signal of models and seismic data.On the basic of this method, unitingthe method of filter, strengthen the weaksignal and predominate reflectance signature of the special geologic bodyclearly. And on the basis of embankment sandstone reservoir feature that issubmerged by background information in ES4 of GAO 8 region inDONGYING sunken, using this method to identify the embankmentsandstone reservoir, to strengthen the feeble signal and predominatethrough-going feature of sands. Secondly, we synthesize some logs (e.g. SPlog, GR log and AC log) to rebuild a new log in the light of the rule of the maxsignal-to-noiseofoutput.Impedance inversion is the main technology of reservoir predictiontechnique. Finally, we process the seismic data and well logs applying themethods above-mentioned and inverse impedance utilizing JASON inversionsoftware. By means of comparison of inversion results, we obtain that theprocessed results can show the divergence feature of reservoir impedancepreferably.更多还原