论文部分内容阅读
在油气藏三维精细描述工作中，常需要对一些三维四变量（Ｘ，Ｙ，Ｈ，Ｚ）地质数据进行分析处理。传统的处理方法多为方位近点距离加权模型和趋势回归模型及克里金模型等。基于现代神经网络信息处理新技术，提出一种简易可行的ＢＰ神经网络三维估值模型，由此可根据研究区域内的已知资料对其未知井点处的储层参数值进行高分辨率估值；采用立体等值线图可视化技术来实现该模型估值的直观图形显示，充分揭示三维地质体某种特征参数的纵横向变化规律。通过对Ｓ气田马家沟组的地层压力等数据的分析处理，结果表明该法是一种有效的高分辨率估值技术，在油气藏描述中值得借鉴使用。
In the three-dimensional fine description of oil and gas reservoirs, it is often necessary to analyze and process some three-dimensional, four-variable (X, Y, H, Z) geological data. The traditional treatment methods are mostly azimuth near point distance weighted model and trend regression model and the kriging model. Based on the new technology of modern neural network information processing, a simple and feasible three-dimensional BP neural network estimation model is proposed. Based on the known data in the study area, the reservoir parameters at unknown well points are evaluated with high resolution The visualization of this model was visualized by using the stereo contour map visualization technology to fully reveal the vertical and horizontal variation of certain characteristic parameters of 3D geological body. Through the analysis and processing of formation pressure and other data of Majiagou formation in S gas field, the result shows that this method is an effective high-resolution estimation technique and is worthy of reference in reservoir description.